WorldCAD Access

Translating textures from one file format to another

It's not so easy

 

by Guillaume Levenq

[This article is excerpted from datakit.com/en/news/textures-one-of-datakit-s-know-how-189.html]

The correct visualization of 3D objects, both in imaging and CAD, requires having, on the one hand, an exact and flawless geometry, and on the other hand colors and/or textures applied with precision on the faces of these objects. 

To put it simply, a texture is a 2D image that is placed as precisely as possible on a 2D or 3D surface of an object, so as to dress it up to make the visualization more realistic. These images can be either simple flat areas of color, or real images representing, for example, a material.

These images most often come from texture libraries, either standard or specific to the editor of the authoring 3D software.

Typical texture library (image source Benchmarq Project Services)

 

These authoring software programs use mapping algorithms, again standard or proprietary, to apply them to the surfaces, which fill the surface by assigning each pixel of the image to a 3D pixel of the surface. These are the so-called procedural textures.

 

From Textures to Renderings

To be usable, this mapping information must be supplemented by lightning information, describing the lighting characteristics of the object -- ambient, diffuse, specular light, and on -- its intensity, as well as the behavior of surfaces with respect to light -- absorption, refraction, reflection, transmission, and so on.

Reflective properties of materials determine their look (image source Cadnav)

 

It is the joint use of the base image with the mapping and lightning information, which constitutes the rendering, giving a realistic rendering of a 3D object, in a given light environment. So much information that it is necessary to retrieve, understand and translate, when we want to transfer a realistic 3D object from one system to another.

An OBJ file, converted to a 3DPDF file with CrossManager in which textures in the PDF file corresponds to those in the OBJ file (image source NASA)

 

How Textures are Converted

When converting from one system to another, the 3D object is recovered, most often in the form of files in B-Rep or mesh formats, with the rendering information that will be applied to it (images, mapping, and lighting).

When converting the 3D object from one format to another, in addition to translating the geometry, it is necessary to calculate from the mapping information the coordinates of the texture on the mesh generated by the tessellation, to apply the image correctly on the 3D, then apply the lightning information to it to have a correct rendering of the object.

Mapping a 2D texture (using u,v coordinates) to a 3D model (image source CJump)

 

If the theory is simple, it is often more complicated in reality. The mapping algorithms used by the various authoring software are almost standard, but with very slight variations from one to another. We have therefore developed our own mapping algorithm, which makes it possible to avoid as much as possible any texture shifts during conversion, regardless of the authoring software, which seems to be perfectly suitable for users of our converters.

But we can also have textures whose format we don't know how to manage. Apart from the standard formats of external images, we can only give information about the presence of an image and the way to get there. In this case, if you want to generate a 3D PDF file, for example, (a format that needs to read the image and embed the associated pixel array) the textured rendering will not be available. On the other hand, for the supported image formats (jpeg, bmp, gif, tiff, png, tga), the 3DPDF will then benefit from the desired realistic rendering, which is very telling, even for occasional users of 3D software.

It will be understood that the conversion of 3D models from one system to another requires a perfect understanding of their geometry, but also of the textures that dress them, in order to make them as realistic as possible for easier interpretation.

[Guillaume Levenq is a development engineer at Datakit]

The best tweets about Google's signon failure

The Top 10

 

Google's signon servers went down earlier today for about an hour, meaning that you could not access its cloud-based software through your account: Gmail, YouTube, Google Drive, Docs, Sheets, Slides, Google News, Google Translate,  Google Ads, Google Domain, Google Cloud Platform Dashboard, Google My Business, and more. Services that don't need an account, like YouTube, Maps, and Search, still worked, if you signed out. Offline software, like Picasa, kept right on working, naturally.

The outage occurred during the night for my timezone, so I never noticed until reading the excited tech media upon waking. I captured some of the best reactions from the twitteratti. As Richard Fernandez reminds us from time to time, Western civilization has become fragile, because it has tossed out too many of the redundancies that would otherwise make us resilient.

As The Register's Tim Anderson put it, "When authentication breaks, everything breaks," and commenter RGP added, "Complexity <> resilience."

- - -

Joe Brown / @joemfbrown: I'm sitting here in the dark in my toddler's room because the light is controlled by @Google Home. Rethinking... a lot right now.

- - -

Jason Jacobs / @jjacobs22: Google outages are the new Snow Day.

- - -

@killedbygoogle: And ya'll want a Google Autonomous car, huh? smgdmfh.

- - -

Harpreet Singh / @harry_jerry: 2020 is so bad it took Google down.

- - -

@iancutress: Annual reminder, while Google is down, that the cloud is just someone else's computer. Always have a backup.

- - -

Ryan Perry / @rynprry: Time to get some Google Candles.

- - -

Dan Gainor / @dangainor: I hear light switches are all the rage.

- - -

Kip Williams / @kiptw: And just think about how Google must feel, not being able to know what room you're in right now, or what you are listening to, or what you are watching.

 

Plus a few political tweets:

Mark Di Stefano / @markdistef: Google's entire suite of products crashing, bringing people to a grinding halt, less than 24 hours before Europe reveals its sweeping new anti-monopoly guidelines. You hate to see it.

- - -

Dan Primack / @danprimack: Well, that's one way for YouTube to prevent election disinformation.

 

 

How to make Eudora 7 work with Gmail servers

Thanks to Hermes

 

Eudora is an old email client. Last updated in 2006, it does not do well on Windows 8/10. It is software that I have been using on my desktop computers since December, 1994, when I first got onto the Internet (from CompuServe). My 96-year-old dad uses it. Today many, who tend to get their email through a Web browser and with free clients like GMail and Outlook, see it as a dinosaur.

 

Eudora's Unique Functions

Except that Eudora does things no modern email client does, and I have tried a dozen alternatives over the years. Here are some of the functions that I cannot live without:

Display two or more mailbox windows at the same time, side by side. This is crucial for me, as my In and Out boxes act as my reminder lists.

• Emails in my Inbox are items I still need to deal with (I color emails that are crucial, like upcoming Webinars, in red)
• Emails in my Outbox are things I am expecting to hear from others, such as outstanding invoices (I color emails that involve income with green)
• All other email is carefully moved to the approximately 100 mail boxes that sort my stuff

With the In and Out boxes side by side, I see at a glance what needs to be done, as well as newly arrived emails. This function is so important that I dedicate one entire monitor to Eudora. (My desktop computer has three monitors.)

Resend Email. I use old emails as templates for new ones. For example, I lead a men's reading group and each week I send out a reminder to the guys. I use the Resend function, which makes an exact copy, which I then modify it as needed. A h-u-g-e productivity boost. (I color emails I tend to resend with brown.)

Queued Send. Some email systems now provide this: the ability to schedule when an email should be sent. I find it useful to write an email days in advance (when I have the time), then queue it to send later -- such as that men's reading group reminder.

 

Eudora's Drawbacks

Being old, Eudora is becoming less and less compatible as Windows changes and the Internet becomes more secure.

Windows Incompatibility. For instance, Eudora expects a 32-bit system, so it only works on 64-bit Windows 7 when you  move some of its files after installation.

It kind of works on Windows 8, and I have given up trying to make it work on Windows 10. Hence, my desktop computer still runs Windows 7 just to accommodate Eudora. (I have a spare Windows 7 machine under the desk should my workhorse give up one sad day.)

Eudora also struggles today with modern HTML code in emails, as well as Unicode extended characters. I can overcome these problems by right-clicking the email, and then choosing Show in Browser.

Internet Incompatibility. Eudora has been working fine with my ISP, Telus, who hadn't done anything over the decades to improve email security. For email access, Telus didn't even require a strong password. But a major lapse last year smartened up the company, when customers went without email for weeks, or, in the case of my dad, months. This month Telus moved the host email system of its business customers (like me) from Dell to Google.

Then Eudora no longer worked. In me, the panic was palpable.

 

The Telus Debacle

Telus shifted my email to Google without particularly informing me. I noticed something was wrong when I wasn't getting any emails throughout the day. I contacted tech support, and slowly it dawned on the techie what happened.

Telus had indeed sent me an email warning me of the switch, but I ignored it. Reason being that Telus allows so much spam through their servers that claim to come from "Telus" insisting I click a button to keep from "losing my account." An email telling me that I needed to take action because I was being "moved to Gmail" was just as worthy of being ignored.

The techie guy said he'd wait a day for the transfer to Gmail to complete and then call me to help me change my email settings. He never did.

In any case, the Gmail settings that Telus lists on its support page are incomplete. 

 

Fixing Eudora

So I was left on my own to fix Eudora. It took about two days. 

My first approach was download and try out a half-dozen email clients. None offered the three crucial features I list above. The "modern" Eudora 8 proved a disappointment, as it is merely a version of Thunderbird, and doesn't do the three crucial tasks.

Thunderbird was helpful in one regard: it does an excellent job of determining automatically the arcane settings for sending (smtp) and receiving (pop/imap) emails. It was thru Thunderbird that I found that Telus posted incomplete GMail settings; eventually I found the full set of settings on Google's support page:

• SMTP and IMAP https://support.google.com/mail/answer/7126229
• POP https://support.google.com/mail/answer/7104828

The problem is, you can't just fix one setting; everything has to be perfect. So here are the steps you need to take:

1. Weaken GMail's security
2. Open up Eudora's access to port numbers
3. Replace aging security DLLs in Eudora
4. Set up properties properly

 

Weaken GMail Security

You  need to weaken GMail's security to allow Eudora access your email stored on Google's severs. The Less secure app access setting is required to allow SMTP, POP, and IMAP access. Here's how to implement it:

1. Sign into your GMail account, and then click the Gear icon, and then choose See All Settings: 

2. Choose the Accounts and Import tab.

3. Next to Change account settings, click Google Account settings.

4. On the left, click Security.

5.   Scroll down to Less secure app access. Enable it. 

6. While in GMail Settings, click on the Forwarding and POP/IMAP tab.

7. Turn on the Enable POP for all mail (even mail that's already been downloaded) setting.

(If you plan to use IMAP, them turn on Enable IMAP.)

8. Click Save Changes.

 

Editing Port Numbers

Regular Eudora does not allow you to change port numbers. To do this, you need to add an extension to Eudora, which fortunately is included in the install but hidden away. Here is how to access it:

1. Close Eudora.
2. Go to the C:\Program Files (x86)\Qualcomm\Eudora Mail 7\extrastuff folder
3. Copy the esoteric.epi to one folder higher (C:\Program Files (x86)\Qualcomm\Eudora Mail 7). EPI files are like extensions to Eudora. This one add more options to the end of the Options dialog box.
4. Start Eudora.
5. From the menu, open Tools | Options, then scroll to the end of the Options dialog box. Notice all the new items that the EPI file added.
6. Click on Ports, and then change the default values, which are meant for unsecure email access, to port numbers used for secure access. See figure below for the numbers to use:

Click OK to close the dialog box.

 

Updating Eudora's Security

Eudora is such an important email client that a few intrepid programmers are working on bringing it up to date. The Hermes Mail project has so far produced files that update the security in Eudora, as required by today's mail servers.  

(Do not confuse Hermes Mail with Hermes Web Mail; the latter is not what you want here.)

The following files solve the problem of Eudora 7.1 not accepting and validating Secure Sockets Layer
certificates -- specifically updating it from TLS 1.0 to TLS 1.2:

1. Close Eudora.
2. Go to sourceforge.net/projects/hermesmail/ and then click Download. This downloads the HermSSL.zip file.
3. Unzip the file.
4. Navigate to the lowermost HermSSL folder
5. First, run vcredist.ese. This is a just-in-case Windows support file that is needed by the new code; if the install returns an error, this means your computer already has the needed support file, so no worries.
6. Copy the following four files to the main Eudora folder, such as C:\Program Files (x86)\Qualcomm\Eudora Mail 7:
    libeay32.dll
   QCSSL.dll
   rootcerts.p7b
   ssleay32.dll
   
   Allow Windows to copy over existing files. 

 

Set Up Properties Properly

There are many options for setting up the Properties of a Personality (as Eudora calls an email account). I have pasted below a screen grab of the ones that work for me. Specifically, take note of these settings:

1. Start Eudora
2. From the menu, choose Tools | Personalities
3. Right-click your personality (email account name), and then choose Properties
4. Enter the settings shown below for Generic Properties and Incoming Mail.
   
   
5. When done changing the settings, click OK to close the Options dialog box.
6. Right-click your personality, and then choose Test. Eudora should report no errors. If it does, then the most likely problem is that you did not change Secure Sockets when Sending to "Required, Alternate Port" in the Options dialog box.
   
7. Now press Ctrl+M to download your waiting email. Life is good again!

 

Generic Properties

User name must be the one that you use to log into the Gmail account; in my case, it is grabowski@telus.net

SMTP server must be smtp.gmail.com

Secure Sockets when Sending must be "Required, Alternate Port". This forces Eudora to (a) use security and (b) use the port numbers you specified in Options | Ports. This is probably the most crucial setting to use in Eudora to make it work with Gmail!

 

Incoming Mail

Server must be pop.gmail.com for POP or else imap.gmail.com for IMAP; for a program like Eudora, you probably want POP service

Secure Sockets when Sending must be "Required, Alternate Port"

 

 

 

From LEDAS: The Polyomino Contest!

Ilya Lichman

 

You may have encountered polyominoes at some point in your life, because they are used in puzzles popular for more than 100 years, including ones like Tetris.

A polyomino is a plane geometric figure formed by joining one or more equal squares, edge to edge. The name was devised by Solomon Golomb in 1953 in a now-classic textbook, and then popularized by Martin Gardner in 1960; see the “Mathematical Games” column in Scientific American.

Today we’ll play with our updated LEDAS logo, which we released a year ago. As you can see, it consists of four pentominoes. A pentomino is a polyomino of order 5, i.e. 5 equal-sized squares connected edge-to-edge.

 

Splitting the LEDAS Logo Into Polyominos

Our question today is this: what is the number of ways in which to cover this area – a 6х6 square, and missing a 2х2 corner by dividing the inner area made of 12 white cells?

Let's agree ahead of time that two configurations are considered the same if they are just made by replacing colors or mirroring the image. For instance, the following two configurations are the same, because we simply repainted a pair of pentominoes and then mirrored the picture:

But what about the white areas in the configuration? As you can see, they make up three parts of the initial 12-square polyomino. Moreover, there are at least two ways to split them:

We have found a way to split the 12-square white figure into three pieces, and then pack all four initial pentominoes and three white polyominoes into the logo area (the square 6х6 missing the 2х2 corner). This means that there are at least three ways to pack the seven polyominoes into the logo’s area. But is there another way? And how many ways are there?

 

The Goal of the LEDAS Polyomino Contest

The goal is to find the best splitting of the white 12-square figure on N pieces. What do we mean by best? It means that we want to have a maximum number of ways to pack everything back into the logo’s area.

For each found configuration, you get +1 point. But for each additional splitting of the white figure, you lose 7 points.
This means that if you find M different configurations after splitting the white figure into N pieces, you have M – 7 * N points. The goal is to maximize your points.

In our example, N=3 and M=3, so the resulting score is 3 – 3 * 7 = – 18.

 

Reporting Solutions

We invite readers to submit their original solutions. Send your configurations to quiz@ledas.com. In your email, please include the explanation and the expected number of points.

We will post the names of all who submit a correct solution. (If you don't want your name made public, then please say so in your submission.) We look forward to seeing what you can come up with!

Visit https://ledas.com/post/869-from-ledas-the-polyomino-contest/

 

 

Canadian gov't tries cracking bitcoin secrecy

Encrypted transparency

 

Canada's tax man, Canada Revenue Agency, is asking the federal court to force Canadian cryptocurrency trading platform Coinsquare to identify every client going back to the start of 2013. CRA says it needs the data to fight tax fraud and the underground economy. It may also be an issue that Canada's left-wing federal government has condemned the country's children and children's children into paying back a $1.5 trillion debt.

Coinsquare, which has 200 thousand accounts,  says "we understand our customers comply with all applicable Canadian laws relating to their cryptocurrency trading activities." The firm, however, had earlier got into trouble with financial regulators, being charged with $2.2 million in sanctions and costs for faking trading volumes, cover ups during the investigation, and firing the whistle blower.

The court has not yet decided on whether the info can be released. 

CRA opened a dedicated cryptocurrency unit in 2018. Tax lawyer David Piccolo says the Coinsquare case "will be a test of the agency’s ability to process the likely massive amounts of data that generally accompanies cryptocurrency transactions." 

Source: nationalpost.com/news/politics/cra-goes-after-client-details-of-major-canadian-cryptocurrency-marketplace-in-battle-against-underground-economy

 

 

How to buy a new laptop

Higher specs, higher price

 

I despair when I see a customer returning a $400 laptop to the store, complaining that it doesn't work well. Or when an organization struggles with its slow-moving laptop that it just purchased. There was a time I was buying a laptop a year for my kids in high school. 

The low-priced ones just aren't worth it.

For a laptop to run well, and to last for years, you need to spend $$$, typically over $1,000. That's a lot of money in an area of life where we had been conditioned to expect ever lower prices with ever higher specs. That that trend made its U-turn years ago. 

My son-in-law wants exactly that: a laptop that runs well and lasts years. The good news is that a well-made laptop runs well for years because computer hardware specs stalled out nearly a decade ago. Other than Apple's untested ARM-based CPUs, nothing Intel or AMD can do will give our computers a 3x speed boost.

The drawback to 3x speedboosts, of course, is that software developers will suck up all that power for themselves. As Bill Gates apparently told his programmers in some earlier decade, assume infinite hardware. That didn't turn out so well.

So, here is how to narrow down the choices among laptops you may want to buy. Begin by expecting to pay $1,000 - $1,500 for it. (My current laptop turns five years old this month. See figure at the end of this article. It is an HP Spectre X360 that cost me $1,400.)

1. First step is to decide on the screen size. This determines all other factors. I prefer small laptops with 13" or 14" screens. The most common size is 15" and there are some with 17" screens, but they are monsters and tend to be rare.
   
   
2.  Pick a CPU type. I recommend at least an Intel i5 for the extra processing oomph (or the equivalent in an AMD CPU). Look for the highest CPU speed in your budget, assuming it's an option.
   
   I tend to go with integrated graphics. The drawback with discrete graphics, like a nVidia board, is that they consume more power and don't give any speed boast to the kinds of software I use -- office productivity, CAD, and Web stuff.
   
   
3. Next, pick the memory. I recommend at least 8GB. 
   
   
4. Choose the hard drive type and size. I recommend you pick one that is all SSD -- solid state drive. There are some hybrid ones, which pair a small SSD with a large HDD, but I want the all-SDD model for speed and for fall resistance.
   
   I suggest a capacity of at least 512GB, because Windows consumes so much disk space for itself, so that smaller drives will just frustrate you. When I got my HP Spectre X360, I picked the smallest drive (256GB), and then waited for SSD prices to fall, then replaced it with a 1TB one. (Not all laptops allow you to replace the hard drive; in any case, swapping out drives is not a trivial exercise.) I like the high-end Samsung SSD drives (EVO or PRO), because they come with caching software that speeds up the drive even further.
   
   
5. Now you can get into the miscellaneous parts:

• Screen resolution: HD (1920 x 1080) is good enough; higher resolutions just make the text too tiny
• I prefer 4:3 aspect ration over 16:9 to get more height, as most software has all their UI elements at the top and bottom
• Touch screen is a must
• Backlit keyboard is a must
• If you can, try out the keyboards in a store. Some laptops have keyboards that are truly awful; most are mediocre
• Touchpens are useless, I find
• You want at least an HDMI out (for a second monitor); some laptops also have a DisplayPort port for a third monitor
• At least three USB ports
• SD card slot
• While USB-C ports give you access to every kind of peripheral, it is annoying to lug along the external port replicator.
• Battery life is voodoo. You can't rely on the vendors' claims, as they are completely hypothetical. As a rough guide, divide their battery duration number by 2. Or 3.
• Spin or Flip or 360. I like laptops whose lids flip all the way around. 

As for brand names, I like Acer and ASUS, because they make laptops for other companies, along with their own brands, and they are located in Taiwan, not China.

TIP: The #1 consumer of battery power is screen brightness. If you need longevity away from a power outlet, turn down the brightness as far as bearable.

My only peeve with the HP Spectre that I use is that the keys on the keyboard are silver, instead of black. This makes it hard to read the letters on the keys, whether backlighting is on or off. So make sure the keys are dark colored. Silver keys look pretty but aren't practical.Silver keys look pretty but aren't practical.

 

 

Spatial 3D Insider Summit

VR only

 

Well, it's 7:00am here on the West Coast of North America, and Spatial 3D Insider Summit should be starting right about now.

I've attended a couple of them before, once in Colorado at at resort so high in the mountains I had a day of splitting headache (from the lack of oxygen) until I got my hands on some tylonol. Then again in Munich.

The Spatial 3D Insider Summit has nothing to do with geographics or mapping. It's the component arm of Dassault Systemes, which'll sell you an ACIS or CGM license for your budding CAD project. Attending the Insider Summit is like attending the OpenDesign Alliance annual conference: you get the head's-up on new technology that (might) be powering your favorite CAD program next year.

For its online conference, Spatial decided to use the VR [virtual reality] gimmick. This requires attendees to download software from a third-party, install it, sign in with the same uid/pw as we registered for the conference, and then take part.

I suppose.

I downloaded the software, but then couldn't be bothered to install it. 

You know, when it comes to online conferences, I am a passive observer -- well, except for the frantic typing. If I have a question, most online conference services have a Questions or Comments panel. The extra effort of being interactive in 3D just isn't my schtick.

I suppose the folks at Spatial thought, "Hey, this is a conference about 3D, let's hold it in 3D!"

Thing is, even in 3D CAD we interact through 2D.

So, no reporting on the Summit from this Insider.

Ralph's thots on Apple's new CPU

A problematic roll-out

 

There is one reason, and one reason only, why Apple is moving to its own, proprietary CPU: to further lock in its user base.

Windows and Linux can run on Apple's computers only because they use Intel CPUs. With Apple's switch to ARM, Mac-labeled laptop and desktop computers become less flexible and so far less desirable. They can no longer provide "the best Windows experience," as some enthusiasts like to put it.

The experience of Google and Microsoft in using ARM CPUs (to replace Intel/AMD CPUs) on laptops is not good. Sure, they get 5G connectivity and longer battery life, but that does not compensate for sluggish operation and, in the case of Windows, incompatibility with important software.

This is being repeated in the Apple world, where third-party software is slow in arriving, especially software that Mac users depend on, such as from Adobe. One CAD vendor that has been candid about Apple's switch is Vectorworks, who are working on making its software work on the new CPU, but has no timeline for completion. Adobe's "next summer" statement is sufficiently vague to allow for delays and worrisome, given third-parties have already had a half-year's headstart on rewriting their programs. Meanwhile, Apple ships the new MacOs for the new Macs now.

Third-party vendors might, of course, be just a little bit ticked off that they need to do all that code re-write just to stay in a sliver of a falling computer market, one in which Macs in 2020 Q2 sold a mere 6.7% of all personal computers worldwide -- down from 7.4% a year earlier.

The telling part is all that Apple left out of the the presentation and the misleading charts that purported to show the necessary superiority of its ARM CPU. Here are some of the complaints from the tech media [with my responses in square brackets]:

• Love charts without X and Y axis numbers. [The charts are meaningless, because there are no numbers on the X and Y axes, and we don't even know if the Y axis starts at 0.]

• Apple failed to show benchmarks of its new M1 CPU against its own Macs running Intel CPUs. [Perhaps the numbers would look less good, and maybe even be catastrophic.]
  
  
• Why only Thunderbolt 3 ports, instead of v4? [I suspect that the new ARM CPU could not handle the bandwidth for video displays required by Intel in the v4 spec, specifically 8K panels.]
  
  
• A 13" computer with 16GB of RAM max should not be allowed to be called a Macbook Pro. The old Mac Minis cap out at 64GB of RAM; the new Mac Minis are limited to 16GB. The new Macbook Pros cap out at 16GB of RAM; the old Intel Macbook Pros cap out at 32GB of RAM. [This could be an addressing issue, in which a limitation of the ARM CPU is the amount of RAM it can address.]
  
  
• They're hedging hard with lots of “first step” and “multi-year” transition so far. [The introduction reminds me of the launch of iPhone, which couldn't do copy'n paste for the first several years.]
  
  
• Craig Federighi is noting how the Mac “instantly” wakes from sleep. [So does my Intel i5-based Chromebook.]
  
  
• No eGPUs. [Some Apple users plugged in external graphics boards, but this is not possible with ARM-based CPUs.]
  
  
• Today's PC laptops have fantastic screens with tiny bezels; face unlock; touchscreens and pencils; 4G modems; tons of awesome features. This looks like a faster MacBook Pro from 2015 with a tiny touchbar nobody wants. [Possiblereasons as to why Windows computers have that 95% market share.]
  
  
• Apple's new MacBook Air and Pro still have 720p webcams. [This could be a supplier issue, or more likely Apple saving money on hardware.]

As Frank Riff put it, "For those who hoped that Apple was going to use the Apple Silicon transition to course-correct on issues such as ports, Touch Bar, and perhaps introduce a new design language or new iOS features such as FaceID, tonight was a bitter disappointment."

Apple bases its claim of a 3.5x speed increase on benchmarks it did not make public, against Windows PCs whose identities it kept secret. This tells me the company has much to fear from the switch once independent benchmarks pop up in the real world.  

I don't like Apple products. I find the hardware limiting (who ships overpriced laptops without touch screens, and crams ports too tightly together? Apple!) and the software irritating (who designs a mobile touch interface with an inconsistent Back button? Apple!). 

The sliding market share of Macs forces Apple to compete against Windows users, rather than for Mac users -- Windows users who  spent the last two decades showing no interest in Apple's fenced-off products.

I admire Apple's gamble on switching to a power-efficient CPU that it manufacturers to its own spec. As I see it, however, the gamble will not pay off.

 

 

Google is a virus

It's not evil if we don't consider it evil

 

Google becoming more aggressive about inserting itself into my world. Here are couple of recent examples.

It has begun demanding that I use my Google credentials to log into Web sites, ones where I already have independent passwords. In the screen grab below, after I am logged into eBay, Google interferes, and in virus-like fashion tried to hijack my login credentials.

Google hijacking a login

One morning, I had an interview with a developer. When he sent a Zoom invite from his Gmail account, Google inserted its own Meet link, as a virus would. See the screen grab.

Google hijacking a Zoom meeting by presenting its own Meet link

As an editor friend said, "I miss the good old days of 'Don’t be evil'." That bromide, about government involvement in our lives, could be rewritten in 2020 as,

"Hello.

I'm from Google.

I'm here to help you."

 

 

Pay people to attend online Webinars

It's happening

 

CAD conferences have done the pivot thing several times in the course of the history of our favorite software. In those early days, there were general CAD conferences, like AEC Systems, where lots of CAD vendors showed up and tens of thousands of users thronged the aisles. Those were delicious times. 

The problem for CAD vendors was that in such an environment customers could see, hear, touch, and learn about competitors.  The stories of the Autodesk booth being mobbed in the mid-1980s are legend -- and true, for I was there. Competitors scowled at the sight that meant the loss of customers.

 

Side Note

I recall speaking with an Intergraph representative upset with the AutoCAD proposition, considered revolutionary at the time: customize CAD yourself. The rep warned me that this was dangerous (or something like that, I don't recall the exact descriptor), and it was much better to have Intergraph do the programming for you.

This was the year, 1986, when a license of AutoCAD went for $2,500 and an Intergraph station went for $50,000 or $1000,000 or more. Intergraph's solution was so expensive because it included the custom hardware Intergraph itself built. AutoCAD's was so cheap because the customer had to buy his own hardware, which in the mid-1980s was actually not cheap. Even though personal computers (running AutoCAD) were much cheaper than minicomputers (running Intergraph), you could end up spending $50,000 on high-end hardware for AutoCAD. Heck, that E-size plotter alone would set you back $18,000. 

Never mind. The times they were undergoing a paradigm shift, and it was exciting to live through it. A different class of people (like me) wanted to assemble their own computers, and also wanted to customize their own CAD systems. Intergraph didn't get it, and for them that was understandable, as they came from the 1970s era of the turn-key mindset, which was normal: one company provided the hardware, software, installation, training, and customization. In the end, Intergraph eventually became another AutoCAD. 

At the time, I was at CADalyst magazine, and we marveled at what our readers were coming up with. My favorite was the stained glass maker who was delighted that AutoCAD could unequally scale his designs to fit customers' windows precisely! A trivial operation for CAD really, but until PC-based CAD came along, otherwise neigh-impossible.

 

Back to the Story

Yah, so CAD vendors scowled at potential customers having easy access to competitors, and so in the early 1990s began holding locked conferences and exhibitions. There was only one message: us. No distractions permitted.

Whereas before anyone with $1,700 for booth space could show up at an AEC Systems show, now even third-party developers who looked like they might be infringing on the mother corp's products were banned. The lockdown extended to stopping outspoken customers at the door, and blocking uncooperative members of the CAD press from attending.

Competitors were reduced to assuming false identities to infiltrate the proceedings (true story), or else set up shop across the street in hotel suites (also true). Some were brazen, buying up billboards on the road from the airport and across from the convention center, even handing out leaflets in the streets, all to convince the faithful of their erroneous belief system. Why a believer would fall for a billboard was beyond me, but hey, it wasn't me who was in charge of that CAD vendor's marketing budget.

Status report from inside: everyone is happy, happy, happy! And so excited. Oh yes. So much excitement!  

Most conferences were not presents from CAD vendors to their loyal customers. Some were free, most charged, and some charged handsomely. Between the conference fee, jacked-up hotel rates, and air fare, customers could be looking at paying $5,000 annually to prove their devotion to billionaire CAD vendors.

Some customers did see an annual trip to Las Vegas (yuck!) as a tax-deductible vacation. Many went for the training, for which they'd have to pay anyhow. And I hear from CAD vendors that they pretty much never made a profit off these events, even the ones who charged handsomely.

 

We're Back to Open

Then the travel ban killed the CAD love-in business. Where  hijackers and shoe bombers failed in decades of trying, the world's tiniest living entity and the world's largest bureaucracies succeeded in just three months.

Fortunately for CAD vendors, the technology for broadcasting conferences over the Internet was already in place, and so we are hearing that 3x and 4x as many customers now take part.

Unfortunately for CAD vendors, the distraction factor is back.

After a 20-year gap, customers again have easy access to a smorgasbord of offerings from competitors. In this environment, CAD vendors can no longer rope in users. In my case, this year I've been to 4x as many conferences as in former times. They're free, they're open, and they take up a few hours instead of a few days. 

So this week, yet another invite arrived for an online conference. This one had a twist. If you attended, you would get $50 off the purchase of the vendor's software, if you signed up early enough. Earlier in the year, PTC offered free popcorn for signing up early, and they made good: I got mine.

It's just a germ of an idea, but as Zoom Fatigue becomes the next psychosis, paying customers to attend could be a clever technique to get the turned-off to once again tune in.

But I do miss the travel to exotic locales.

 

Seven tips for working with Substack

Hidden and not so hidden functions

 

How to Preview Your Newsletter

SubStack's process for previewing your newsletter is a bit awkward:

1. Click Settings at the bottom of the screen.

2. In the Post Settings screen, scroll down until you reach Secret Draft Link. 

3. Click Copy Link.

4. Paste the link into the address bar of a Web browser. 

5. To exit the Post Settings screen, click the X in the upper right corner.

 

How To Move a Paragraph

Hold down the Ctrl key and then double-click the paragraph. Notice the blue border that surrounds the paragraph. Drag the paragraph to a new location.

How To Add Captions to Images

Click the image. In the upper-right corner of the image, notice the big blue dot with three little white dots.

1. Click the blue dot.
2. Select Edit Caption.
3. Below the image, replace "Caption..." with your caption text.

To distinguish caption text from regular text, SubStack centers it and renders it in a small sans serif font.

When the Newsletter Gets Too long

Although SubStack claims no limits to their service, there is in fact a limit -- the length of the newsletter in bytes. I sometimes produce long-form articles, and eventually this orange warning alert appears at the bottom of the screen:

Click the tiny i button to learn why:

Gmail will truncate emails that are more than 102KB. We display a warning after you've exceeded Gmail's size limit.

The 102KB limit includes text and images. When you have more images, then less text is allowed. To eliminate the warning, remove one image at at time, or else cut and paste text for use in another issue of the newsletter.

 

Working on Multiple Newsletters

I often work on two or more newsletters at a time. I keep each one open in its own tab of a Web browser. In other tabs, I keep a copy of my newsletter's template and the dashboard. In this way, I can add text to any draft newsletter as I come across new and interesting information.

Create a Template for your Newsletter

SubStack does an excellent job when you copy text-figures from one newsletter to another. It does not lose any formatting.

So that I don't have to type things repetitively, I have created a template for my newsletter, which is just a draft one that I have named "Template." See the screen grab below for some of the elements I have in my template. When I start a new edition of my newsletter, I copy the entire template and paste it into a New Post.

Best Time to Send Out a Newsletter

This is a bit of info I learned when I still was with MailChimp. It does an analysis of email opening times to determine the best time to send out your newsletter. "Best" is defined as getting the most opens for each issue. For me, it was 7am in my time zone, Pacific.

To set the time for your newsletter:

1. Click Publish at the bottom of the screen.
2. Click Schedule for Later.
3. Choose the date and time at which to have SubStack send it out.

 

Now is the time to cancel MailChimp

"In our sole discretion"

 

About five years ago I was looking around for the mass email service to take over my homegrown effort. I was spending too much time updating the mailing list, and there were firms who promised to take the work over for me -- for a fee, 'natch. I settled on MailChimp, and for the most part it worked mostly to my satisfaction, at first.

I have just two needs in sending out my weekly upFront.eZine newsletter:

1. Take care of additions, subtractions, and changes to the mailing list
2. Let me create my newsletter in the way I want it to look (one of my other jobs is book designer, so I am picky in how text looks)

Once the relief of no longer having to maintain the mailing list began to wear off, however, I realized just how much convenience was costing me. Translated into my local currency, it came to about $960 a year.

Like most similar services, MailChimp charges on the basis of how many subscribers you have on a mailing list. To reduce my cost, I began manually paring down the list, removing those names MailChimp found never opened the newsletter. Over time, I got the list down to the lowest price level, costing me about $600/year.

No sooner had I cut my cost to the lowest level possible, MailChimp increased the price by about $120/year. Their excuse was that they were adding so many new functions -- none of which I would ever use, nor even asked for. That both events happened in the same week so irritated me that I began looking for alternatives, but found none that satisfied me.

 

We're All in This Together, Except When We're Not

During the summer of COVID-19, MailChimp hit its customers with a one-two punch. First, it made a massive change to its user interface that reduced the names of major functions to inscrutable icons. That was the next breaking point for me. I should not, as a customer, to be required memorize the meanings of icons.

This move highlights one of the primary drawbacks to working with cloud-based software: customers cannot freeze releases to stay with a version that works for them best. This is the reason some of the desktop software upon which my business depends is frozen at releases that came out in 1998, 2000, 2007, and so on. Visio 2000 works just fine, as do Eudora 7 (2007), PaintShop Pro 6 (2000), and Word 2007. Perhaps that was the golden era of software.

Another drawback to cloud-based software (of which I use little, one of the exceptions being this blogging service) is the hesitation of speed. I was one of those who endured the early PC era, where everything seemed painfully slow. The cloud returns us to the 1980s, where I am forced to wait for a screen to change, wait for a file to save, wait for text to update, wait for a preview to show up. Pretty much every interaction is click -- wait -- click -- wait... MailChimp, in trying perhaps to be all things to all marketers, seems to have a lot of code slowing it down.

By late summer MailChimp was piling up demerit points from me. Getting more expensive, inscrutable interface, slow reaction time.

The final straw that broke my camel-like patience was a second price increase that arrived in the early fall, even as the second wave of COVID-19 hit the world. We truly are not all in this together. Again, the same lame excuse that they were adding all kinds of new functions for which I have no use. I was now up to paying about $800 a year to send out my free newsletter on their cheapest plan.

Ratcheting up the cost

 

Sendinblue

Over a period of 2-3 weeks, I signed up for a half-dozen mass emailing services. A couple of them I got seriously involved with, trying to make them work with my newsletter.

Hot tip: All of them import email lists from competitors (usually through CSV format), so remember to export your list before cancelling a service. Some also will import your newsletters.

One, Sendinblue from Paris, was promising in that their annual price was $300 -- about 2/3 cheaper than MailChimp. Tech support was wonderfully responsive. I paid the fee.

But their editor was bug-ridden, including a very weird bug where sometimes it would add a line of JavaScript on its own, and then announce it could not save the document -- because it contained JavaScript. I reported it, and the techies fixed it in a few days. (Even weirder, the JavaScript pointed to a legitimate Web site, making me wonder if a programmer had been copying'n pasting code.)

But it was the second major bug that killed Sendinblue for me: It would not upload most JPG and PNG images to my newsletter. It complained, "File format not valid." That bug the techies never solved by the time I cancelled my account.

Two upFront.eZine newsletters went out with Sendinblue, and this is why they lacked images to illustrate articles. Remarkably, and to my relief, they refunded my fee, pro-rated, so I lost only about $10 on the experience.

 

Constant Contact

Another one I spent time with was MailChimp's largest competitor (I think), Constant Contact. Their big advantage was (or so I thought) that they cost almost as little as Sendinblue. They automatically offer free telephone support to get new customers going. Very lux.

Again, the problem was with images. The live tech support person spent several phone calls with me to figure out my situation, which, on the face of it, is so simple: click a spot anywhere in the newsletter, insert an image file from my computer's hard drive. Both MailChimp and TypePad (this blogging service) do it, so I was left puzzled why so many mass mailing services I tested WON'T do it. Constant Contact had such a weird multi-step workaround that I can't even describe it. Even more weirdly, the techie told me Constant Contact had no plans to repair the problem. This was disheartening to hear.

The entire time I was telling other CAD editors about my experience, and one of them warned me that Constant Contact was known for being very expensive, a claim I found puzzling. When I first signed up, I told Constant Contact I had under 5,000 subscribers, and the fee quoted to me worked out to about $350/year. Some days later, I went to check their range of plans, and the quoted fee shot up to something like $1,200 a year. Yikes! This made MailChimp, in comparison, cheap. The huge change in price made me pretty angry. All that time wasted.

The reason for the Constant Contact name is the constant contact from their sales team. At the end, my contact hung up on me in disgust, as I seemed incapable of grasping all the wonderful features they would offer for their exorbitant fee. Never mind that I didn't need them. The customer is always wrong. 

Fun fact: All of these mailing services stringently check that your mailings could not be considered spam. Indeed, two services refused to take me as a customer because the domain of my email address (telus.net) did not match the domain name of my newsletter (upfrontezine.com). Some, however, effectively sent me spam, as they left out the legally-required Unsubscribe button in their email communications with me, even after I cancelled my accounts with them.

 

Substack

Frustrated, I began looking at the mass emailers were being used by the emails coming in to me. One was Substack. I remembered I had a Substack account, which I had signed up for it at some point, but had never investigated further. In desperation, I tried it. It offered the two things I needed most, at the price I wanted most: free. They even allow ads in newsletters. They make money by charging 10% if you set up a subscription fee through their system. 

The formatting is very limited; there isn't even a way to center text! At least they let me punch in an image anywhere I need one. 

Substack's simplistic editor

So, in the end, after checking out a half-dozen mailing services, upFront.eZine now goes out with Substack. I do worry, however, what this venture-funded service might change in the future.

 

MailChimp Goes Woke

Then  the news came out this week that MailChimp changed the Standard Terms of Service. It had decided to go all Twitter on its customers.

Twitter smothering its customers with mother-knows-best

MailChimp added the paragraph that I highlighted in yellow in the screengrab below, in which it promises to read all the emails you send out and block those that -- in its sole discretion -- it does not like. "Sole discretion" means no appeals:

• It does not describe how the censoring process will take place
• It does not define the word "important," but I think we can guess what it refers to 

As a publisher, I am hard-core pro-free speech. Just in time, I had escaped MailChimp. It may be worthwhile for others -- in our sole discretion -- to cancel MailChimp. There is, of course, no guarantee that Substack will escape the attention of the Woke brigades, but for now it remains a sanctuary site.

MailChimp's desire to be the Twitter of your newsletters

 

Who Had the First DWG Editor?

Felix CAD on first

 

I always thought Softsource's VDraft was the first CAD program to use DWG as its native file format, but it looks like Graebert beat it with Felix CAD by about a half-year or year.

Digging through the historical record, it appears that Felix CAD was first (1995) and VDraft second (1996). Both were developed independently from Autodesk and from each other. Felix CAD in Germany, after Autodesk took away the European AutoCAD distributorship from Graebert, and VDraft in the USA based on a DWG/DXF plug-in it had written earlier for Web browsers. I ended up writing documentation about both CAD programs.

Early promotional slide for Felix CAD

 

The history of Graebert reads, "The first release of Felix CAD in 1995 was 100% written in-house, including the code to read/write DWG files." Felix CAD today is known as ARES Commander.

We here in North America didn't know about Felix CAD, which is why, when I wrote an email in January 1996, I said, "The February issue [of the CAD++ newsletter] will feature the premier article on SoftSource's new $500 CAD software that uses AutoCAD DWG format as its native format -- the first AutoCAD clone."

VDraft boasted multiple windows for editing DWG files, which AutoCAD, at the time, could not do

The first knowledge I had about Felix CAD came in an email from March 1996 written by the German translator of my CAD++ newsletter. He reports interest from "Mr. Stahnke of Felix GmbH." (Mr Stahnke is still with Graebert today, continuing in his role as OEM and Partner Manager.) The Web site at the time was fcad.com, which now is hosted by "Fine Chemical Application Development Corporation (commonly referred as FCAD) is the world's leading fine chemical company."

 

The Distinction

I need to note that Felix CAD was not the first non-Autodesk product to read and write DWG files. There was a distinction back then, as there is today:

DWG Compatible. Some CAD programs can read/write DWG files, translating them to and from their own native format. In this regard, I believe that Microstation from Bentley Systems was the first DWG-compatible program.

DWG Based. Other CAD programs use DWG as their native file format, a much tougher programming job, and that is what we are talking about here.

 

Garage DWG

Then the market exploded. All kinds of independent developers were writing APIs that could read, display, and write DWG files. In August 1997, Roopinder Tara and I talked about me writing a comparative review of DWG-compatible software for Cadence magazine. Our initial list consisted of

• Visio Technical
• Visio IntelliCAD (if a beta is available)
• SoftSource Vdraft
• Intergraph Imagineer
• IMSI TurboCAD
• Bentley MicroStation
• AutoCAD LT
• FelixCAD
• DynaCAD
• CADkey
• Corel VisualCAD

(Those in boldface use DWG as their native file format; the others could read and/or write DWG files to/from their native format.)

Today I count over 30 programs using DWG as the native file format. Most are based on IntelliCAD, some on Graebert's ARES, a very few are developed independently, but all use DWG read-write libraries from the OpenDWG Alliance (now Open Design Alliance), which was launched by Visio in late 1998.

The First First

If we want to talk about the really first in DWG, then we need to jump back to 1988, when a Dutch company, Cyco Automation, released the first ever DWG viewer. They had cracked the code in how to parse DWG files and then display the result in software that did not rely on Autodesk. 

Packaging for the first release of AutoManager, showing its DOS interface

 

Their DWG viewer, called AutoManager, was such a breakthrough that we didn't understand the point of it, at the time. For us, working at CADalyst magazine, we would just fire up AutoCAD if we wanted to see the content of a DWG file. This was under MS-DOS, where you could (for the most part) run only one program at a time. Choosing between AutoManager or AutoCAD as to which program to run, it just made sense to start AutoCAD, as it did more than just view DWG files. 

Then Autodesk went to war over DWG, but that's a whole 'nother story.

 

 

It's Hard Buying a Chromebox

A rare beast

 

For nearly ten years, I've had a Gigabyte microPC in my home's entertainment room, whose job is basic: view movies from Acorn, Britbox, and so on. It runs a Web browser that does all the work streaming us movies.

Initially, I had used an Internet-enabled DVD player, which showed Netflix and other services, as well as played DVDs. But these so-called "smart devices" have short lifespans when it comes to staying up to date and still being able to access Netflix and its ilk. This is something the salesman will never tell you.

Anyhoo, I figured getting a microPC running Windows 10 (free upgrades for life!) would solve the long-term incompatibility problem. It is a minimal system, just 4"x4"  in size, hosting an i3 CPU with 4GB RAM and 64GB SSD. But that's good enough for running a Web browser.

The offending miniPC

 

Fake Audio Driver

But then problems ensued, one annoying, the other really annoying.

A couple of years ago, the sound disappeared. It took me a while to track down the problem: the audio driver had been replaced. Eventually, I found the solution in downloading the OEM sound driver, and installing it.

The problem, however, became more and more regular. Something -- Windows, I suppose -- at random intervals replaces the Intel-branded audio driver with another one. My wife would wait patiently as I did the reinstall of the correct driver, a process taking 6 or 7 minutes each time. The problem became regular, first every few weeks, and then more recently every couple of days.

Reading online, I found that this is a known problem: Windows 10 arbitrarily replaces perfectly good audio drivers with non-working ones. I tried a lot of the solutions that were suggested, but none were permanent.

 

CPU Cycle Eater

The second problem emerged from the muck a couple of months ago. The miniPC would grind to a crawl.

Checking the Task Manager, I could see that the CPU was running at 100% -- right up to the moment I opened the Task Manager. I never did find what was sucking up the cycles, as no virus haunts the machine.

Between not being able to hear movies and movies grinding to a stop, something else needed to be done.

 

The Chromebox Solution?

So last night I made the executive decision to replace the miniPC with a Chromebox. This, I hope, will solve the evil audio driver problem and the even more evil CPU crawl problem. I love the Asus Chromebook Spin 13 laptop I bought a year ago, and so hoped to re-experience the experience.

That was when I discovered how hard it is to buy a Chromebox, if, as in my case, you want the following specs:

• Reasonable price -- its primary job is to merely run a Web browser, after all
• Reasonable specs -- enough to run a Web browser
• Long term ChromeOS support -- well, yah
• Runs Android apps -- nice to have

Firstly, I discovered Chromeboxes tend to have horrendous prices, some going for nearly $1,600 here in Canada at Best Buy.

Regularly $1,586, after tax

 

The prices are especially horrendous especially when you consider the boxes come with no hi-res touch screen, no backlit keyboard with no integrated trackpad and no stylus. I didn't want to spend as much as my Acer Chromebook, which was $850, for something that was so unfeature-ladden.

Then I discovered how rare they are, as compared to Chromebooks. One of Google's lists holds 242 Chromebook models, but only 22 Chromeboxes. And maybe that's why the boxes are expensive -- minimal competition.

So I found a reconditioned one on eBay with a i7 CPU for US$140, before shipping. I ordered it.

 

Android Support

Then a though struck me. I want one that runs Android apps. Does this one? Short answer, no.

Here is Google's page listing the devices that support Android: https://sites.google.com/a/chromium.org/dev/chromium-os/chrome-os-systems-supporting-android-apps.

 

LTS

LTS is industry jargon for "long-term support." It's most commonly seen with the Linux operating system, where a release is supported by the distributor for a few years. It also happens in ChromeOS but in a different way:

Google says they will provide ChomeOS with updates during several years for specific models of Chromebooks and Chromeboxes. The newest models are supported until 2029, while older ones are already dead, upgrade-wise.

The official list is here: https://support.google.com/chrome/a/answer/6220366?hl=en.  That reconditioned one I found on eBay had got its last update in 2019. Fortunately, I was able to cancel the order.

 

Which? Chromebox

So now the list of Chromeboxes for me to consider was considerably narrowed down, if I wanted Android support and LTS ChromeOS support -- just seven models. All of these have LTS of 2025:

1. Acer Chromebox CXI3
2. Aopen Chromebox Commercial 2
3. ASUS Chromebox 3 CN65
4. CTL Chromebox CBx1
5. HP Chromebox Enterprise G2 
6. Promethean Chromebox
7. Viewsonic NMP660 Chromebox

The list looks slim, but isn't. The Acer CXI3, for instance, comes in nine hardware configurations in the USA, four in Canada with prices that range from $320 to $1,000.

In the end I decided on the bottom-of-the-line CXI3 model, with these specs. In brackets I show the specs of the miniPC:

• 1.80GHz Intel Celeron Dual-core CPU (1.7GHz i3)
• Intel HD Graphics 610 (similar)
• 4 GB RAM (same)
• 32GB SSD (64GB)

I'm interested to see how it turns out.

Online Event Fatigue

No immediacy, no intimacy, no humanity

 

One of the benefits of online conferences is that they cost corporations much, much less to mount. Instead of one annual physical conference that deals with everything companies want to cover, they now mount frequent mini-conferences. This is gone to the extreme with phone vendors, who now have a separate launch event online for each new model as each chrysalis emerges from its factory throughout the year. Some CAD vendors engage in this multi-event approach, too.

The result is event fatigue, in which the intended audience, after a while, says, "I really don't care." 

Online events are like local events. On the day of a local event, say a ten-minute drive away, I'd have to push myself to get in the car and go there. Online events are the same. When they are a ten-second GoToMeeting download away, they can prove to be a irritant, an interruption to the daily schedule. The difference is that most online events are archived, so we could view them later at our leisure -- kind of like we'd record programs on our VCRs and then never watch them later.

It's the immediacy that matters.

Also, it's intimacy that matters. Being physically with other people at a conference site, making happenstance get-togethers, socializing over quiet dinners and too-loud parties, bumping into vaguely-remembered faces ("Hey, how are you!"), catching the eye of a would-be contact, exploring the exotic locale in which the conference is located (Las Vegas, of course, excepted). 

I've probably been using GoToMeeting and Skype for business meetings for a decade now. I like their efficiency. After a half-year of multi-person Zoom meetings, I now despise them. And I think that the keyword is "multi." For it is the indiscriminate equality of the 33 people -- no, not people, faces -- that makes multi-face online meetings disorientating.

It takes just one cough for a minor participant to become the star of the event, albeit for seconds, as Zoom zooms the face full-size and highlights it with that bright yellow border. I've seen a participant folding her laundry, and so we know what colors of panty she prefers. Another wandered to the side to get a snack and then returned to the Webcam to thoughtfully pick his teeth clean. It's disconcerting.

The thing about annual in-person events is that they feet important. It's the one time this is happening, for an entire year. And so they made us feel important just by attending, along with all that the complex pre-trip planning heightening the anticipation -- best airfares! hotel locations! cheapest ground transportation! foreign currency! local SIM cards! how big a backpack will I sneak aboard the airplane!

Then there is the discontinuity of the live versus canned online conference. I am in front of the computer screen live, but the presentation was filmed days before, and edited to perfection. I feel cheated. At least live online conferences exude some feeling of humanity .

That's the key, and the evil inherent in the coronavius and resultant disruptions. By disconnecting ourselves as humans, by masking our smiles and grimaces, by avoiding each other on the sidewalk, we become automatons -- full autonomous, fully anti-human. When we fight against the virus, even those who do not believe in evil are nevertheless fighting against evil intent on destroying our interconnectedness. It is only something humans experience when two arms brush accidentally against each other on that shared aircraft arm rest.

Quite frankly, I enjoyed the year-long break from overseas travel, but now I am ready to return. The first in-person CAD conference for 2021 is Hexagon's (in Las Vegas, yuck!). The CTO of another CAD vendor told me he would wait another six months before committed funds to his otherwise-annual in-person conference. Open Design Alliance said it will hold its annual conference next year in-person in Munich (hurrah!). So, there still is hope for humanity to re-emerge from the lock-downs mandated by politicians-cum-scientists.

All these conferences planned for 2021 could, of course, be cancelled. 

 

How USB-C Headphones Differ

Digital clarity

 

In I Come to Praise USB-C I explored how the now-ubiquitous USB-C connector single-handedly handles nearly everything for which computers and phones formerly needed multiple ports -- memory cards, regular USB ports (for printers, external hard drives, camera, etc), ethernet, displays, and so on. Bulky, but doable. See figure below.

All ports for the price of one

The bonus for phone and laptop makers is that USB-C lets them cut costs by providing fewer ports, even as they raise the prices on us. 

The other peripheral that the USB-C port on your phone might handle is a set of headphones. Headphones over USB-C is a different matter from all the other peripherals I mentioned. Those peripherals are either all-digital or they handle some aspect of rasterization internally (such as inside cameras and printers). None of them is analog; you might have noticed that none of these port extenders have 3.5mm headphone jacks.

 

Vibrations are Good

Analog is what our ears need. When we listen to a string quartet or a jazz trio, our ears are interpreting the waves of sound produced by vibrating instruments. The cellist drawing her bow across the strings to vibrate them; striking padded hammers on a grand piano's strings vibrates them; a saxophonist blowing into the mouthpiece vibrates the reed; a trumpet player buzzes his lips against the mouthpiece.

Vibrations being carried through the air, from the instruments to our ears, vibrate our ear drums, then the inner ear bones, which stimulate nerves into sending signals to the brain, which interprets the signals as music, to our listening pleasure.

Analog means that the sound our ears receive is analogous to what the instrument produced. Digital means that the analog sound is digitized up into pieces, commonly 44,100 pieces per second of music. CDs, music streaming, and MP3 files all store music digitally.

(Digital instruments like electric pianos and syndrums convert their digital signals into analog ones, and so generate the vibrations after the fact; the quality of the signal and of the conversion determines the quality of the sound made by these artificial instruments.)

There is no such thing as a digital speaker. All speakers and headphones are analog. All use the same technology that was figured out more than 100 years ago: the analog signal is an electrical voltage that varies, stimulating a magnet to move a diaphragm rapidly in and out, producing sound waves that our ears pick up as vibrations.

(Microphones are the exact reverse of loudspeakers: sound waves stimulate a diaphragm, which moves the magnet, which generates the electrical voltage, which is stored as sound.)

Yup, when a earphone vendor tries to impress customers with that dramatic image of the exploded view of the innards, you are seeing, amongst all the clutter, a magnet and a diaphragm -- no different than from 1915.

Magnet, meet diaphragm

 

About DACs

The reverse occurs inside computers and portable devices. We store entertainment in digital formats, such as MP3 (for most music), AAC (for music purchased from Apple), and MP4 (for movies). For us to hear the music, the computer has to convert the digital signal into analog sound. In every single case, this is done by a small chip called a DAC -- digital-to-analog converter. 

It is the quality of the DAC chip that determines the quality of the music you hear. High-end MP3 players have two or more expensive DACs that do a good job; phones might carry cheap DACs to do a so-so job. If the customer is unconcerned by the quality of music, then the cheap ones are good enough.

When it comes to headphones plugged into the USB-C port, the DAC has to be relocated, because the phone is delivering sound in digital form right up into the port. Now it is the headphones that have to hold the DAC chip to covert the digital signal entering through the USB-C connector into analog sound for the headphones' miniature speakers. 

In a USB-C headphone, the DAC chip is stored in the USB-C connector, the part you push and pull into and out of the phone. See figure below.

Where the DAC chip hides

The same occurs with headphones that connect over Bluetooth. The laptop or phone sends digital music through the wireless Bluetooth connection to the headphones (or speakers), which then have the job of converting the digital signal into analog sound. The drawback to Bluetooth is that music, already compressed into MP3 format, is further compressed to fit the limited capabilities of the Bluetooth data transmission specification.

Those phones without the 3.5mm analog port for headphones get to also eliminate the DAC chip -- a double savings for companies like Apple. For these kinds of limited-function phones, you can get a USB-C to 3.5mm adapter that lets you use traditional headphones. The adapter carries the DAC chip inside; some adapters have both 3.5mm and USB-C at the end, so that you can charge the phone while listening to it. See figure below. It's awkward, doable, and more cost-convenience that's been transferred from the vendor to the consumer. 

Power and sound

 

USB-C Headphones

To see if USB-C headphones are any good, I had bought a couple of cheap $5 ones from eBay last year. They truly were awful. The sound was badly distorted, indicating very low quality DAC chips. I guess that was to be expected.

This year, I tried again. This time I got a moderately-priced set for $40 from Amazon, Dyplay, which had decent reviews. I picked thise ones in part because of the nub; I need such nubs to keep the headphones in my ear; without a nub, they fall out. See figure below. 

Dyplay UCS-C headphones; also available in white

The difference in sound was remarkable. It was excellent.

Now, know that the music player on my Android phone is Neutron, which produces the best sound  through its own 64-bit audio engine. It is aggressively updated, so I was only mildly surprised when Neutron knew I had plugged in a pair of USB-C headphones. This meant it would instruct Android to ignore its own DACs and direct the signal directly to the USB-C port.

These headphones claim to have ANC -- active noise cancellation -- which I found is minimal. I would call it PNS -- passive noise suppression -- which is a design of earphone that plugs into the ear canal so well that outside noise is effectively reduced. I have a couple of pairs of such PNS earphones that my wife and I use on airplanes -- much more compact than those bulkly Bose-style ANC over-the-ear headphones that need batteries!

Anyhow, the point is that USB-C headphones don't need batteries to power ANC, because the USB-C port provides the power. 

I am fortunate that my smartphone, the Asus Zenfone 6, handles all forms of music output:

• 3.5mm jack
• USB-C port
• Bluetooth
• Loudspeaker

 

 

From Cassette to MP3

Limitations galore

For my daughter's upcoming major birthday, her husband wanted to make a playlist of the music she used in her figure skating routines. It was my job to hunt them down.

One song was on a CD, so that was easy to convert to MP3. There is lots of free software that does that, even iTunes will do it once you jump through the correct hurdles.

Another song was on a cassette tape. There was a time when I had hundreds of tapes, which I listened to, to spare wear and tear on the roughly 400 vinyl records I had bought. There were also many tapes I made from radio stations around the world, from the mid-70s to the mid-90s, and so there was a time a couple of decades ago when I converted all those records and all the tapes to MP3 files.

In the case of my daughter's figure skating piece, this meant remembering two decades later how to covert music from a cassette tape to an MP3 file. It takes the correct hardware and the right software.

 

The Correct Hardware

I dug out my Pioneer cassette deck from storage, and was pleased to find that it still had attached the RCA-3.5mm cord that I needed.

RCA-3.5mm cable

The RCA ends plug into the back of the cassette deck, into the LINE OUTPUT ports, one each for the left and right channels. 

RCA plugs attached to a cassette deck's Line Output (Play) ports

At the other end is the 3.5mm plug, which goes into the computer. Not just any computer nor any port. It needs to be a computer that has a Line In port. Most today lack it, as most have a combo plug that handles microphones and headphones in a single 3.5mm port. An older laptop of mine had two ports, one for headphones and the other one for a microphone.

3.5mm plug attached to a computer's pink Line In port

You cannot use any of these with a cassette deck, as the output volume from the deck is too loud, and you will record mere noise -- distortion, actually. Microphones have a much quieter output than the Line output from a stereo component, and so the computer boosts the volume for what it thinks is microphone-level sound coming in.

Fortunately, my decade-old desktop computer has all three 3.5mm ports -- line in (pink), microphone (blue), and speakers/headphones (green). Hurrah for old hardware.

 

The Right Software

You need software that reads the sound coming into the Line In port, and then converts it to an MP3 file.

In the past, I used Quintessential Music Player, aka QMP. It has all kinds of wonderful tricks built into it, but is no longer supported. For instance, it can read a stream from an online radio station, separate each song out of the stream, and then give each song a filename made from the musician-album-title.

I fired up QMP, and set it up to read from the Line In port:

Accessing the Rip & Encode window

1. Click the small red dot in the main window of QMP.
2. In the Rip & Encode window, click the Encoding Source button, and then choose Record from Line-In.

Selecting the input from which to record

Make sure that the recording will be saved in MP3 format:

1. From the File menu, choose Preferences, and then choose Ripping and Encoding. 
2. Under the Select Encoding Format, choose MP3 Audio. 
3. Check that the Lame encoder file is available at Using Lame Encoding Engine.
4. Click Save and Close. 

Settings in QMP to record in MP3 format

But it complained that the MP3 encoder, called Lame, was invalid. Huh.

I searched around the Internet for Lame, but had a hard time tracking it down. (Lame is an open source MP3 encoder, whose name is an acronym for "Lame Aint an MP3 Encoder," a self-referential acronym typical among ironic Unix programmers.) I knew I had QMP installed on an older laptop, but when I fired it up, QMP complained of the same problem. Huh.

By chance, I found I had another MP3 converter program on the laptop, and it had the Lame DLL (dynamic link library) that I needed. I used Gmail to send the file to myself. Google, in the infinite wisdom of its A.I., said, "Nope." Huh.

I encountered this blockage before. For me, the easiest way to copy a file from one computer to another is to simply email it to myself. Lately, however, Google has been blocking me from sending myself files it considers dangerous, such as DLLs and APKs. Google, after all, knows better than me what is good for me. If level of censorship represents the future of computing, I want no part of it.

So I used my non-Google email account, and it said the same thing: Google would not let me send a safe file to myself. Then I recalled that my non-Google email provider had recently switched to using Gmail as its email engine. Sigh.

The workaround, of course, is to use something like Dropbox, or PCloud, in my case. The Lame file transferred between computers successfully. (Thank you PCLoud for not censoring me.) Through the Ripping and Encoding preferences dialog box, I told QMP where to find the Lame file, and it was happy.

 

The Two Steps

There is no automation for the actual recording activity. It takes two steps:

1. Press the Play button on the cassette deck, and then click Encoding Start on QMP.
2. When the music is over, click Encoding Stop on QMP.

Clicking "record" on QMP

Converting the two-minute song took about me about two hours.

 

  

Firing up my iPod Touch, ten years later

Model MC011C

 

As the Palm Pilot era was beginning to wane, I began to look for a replacement. I had bought my first Palm Pilot, the original, in 1996, and then bought nearly every successive generation after I found it to be the most productive tool for me -- ever. Its simplicity meant it was efficient, and I demand efficiency.

The iPad Touch from Apple was enticing, and so in early 2010 I bought one of the third-gen models. I knew from experience that it was best to buy a device with the most available storage RAM, and so I got the 64GB model. It was not cheap (Apple charged me an extra $100 for the added 32GB RAM), and it lacked a camera and GPS, despite having the incongruous Photos and Maps apps. I never did fill up the 64GB, and so could have saved me a hundred bucks.

Nevertheless, that first night I was entranced by what a modern, WiFi-enabled, touchscreen device could do. I was up until 1am exploring it.

It was at the same time remarkable and limiting. I found the closed Apple ecosystem, lauded non-stop by its compatriots in the tech media, stifling, me having been spoiled by the richness of Palm third-party development. The lack of a camera on the iTouch was dismaying, and I found the skeuomorphic GUI just awful.

Eventually, I gave it to my daughter for her time studying in England, and later that same year I bought the first Galaxy phone from Samsung, the Android-based Galaxy S I9000. Its GPS never worked properly, but it otherwise proved that Android was the open platform for me. 

 

Ten Years After

Yesterday I hauled the rather fat and tiny iPod Touch 3rd Gen out of the drawer and charged it up. The aluminum back is so scratched that I can no longer read the small text on it, and it took me a while to track down which model it is.

Aluminum back, after a decade of scratches

It runs iOS v5 and most apps still work. Exceptions are apps that rely on external providers, such as the Weather, YouTube, and Newstand apps. The iTunes app still works, and for a device meant to play music, it is really the only one that matters. I found that I could download songs I purchased on iTunes from over the decade. It runs slowly, but not frustratingly so. 

Downloading purchased tunes from iTunes

I connected its old Bluetooth (buried under Settings | General | Bluetooth) to my new Bowers and Wilkens P7 headphones. Listening to acoustic jazz albums like Ralph Towner's Solstice and Steve Kuhn's Pastorale reveals that the Touch offers up remarkable sound, albeit with the occasional hicough as the high-bitrate AAC data stream overwhelms the 600MHz CPU.

Even though it introduced me to the post-Palm world of modern mobile devices, the iPod Touch was for me, in the end, never more than a curiosity. 

Standards for collecting point cloud data

3" or 1/4"

 

3D point cloud data (nothing to do with storage or computing clouds) recorded by high-speed laser measuring devices are the new norm for data collection, inside and out. They are used to produce large-size scans of roads, buildings, and manufacturing plants. Versions are available also for small size scans, such as of just-manufactured parts and car bodies. 

The problem of raster has dogged it since it was invented: the higher the resolution (ie, more accuracy), the bigger the file size. For 2D images, like photograph, the increased file size is by the square: twice the resolution make the file size 4x bigger. Compression schemes, such as ZIP and JPEG, reduced sizes by about 10x.

For 3D laser scans, the problem is cubed: doubling the resolution makes the file size 8x bigger (2x high by 2x wide by 2x deep). While storage space is dirt cheap these days, and new algorithms tackle the problem of billion-point files, there still is the time it takes to collect points. So there is an interest in balancing resolution against file size. 

Defining Accuracy

The Massachusetts Association of Land Surveyors & Civil Engineers (MALSCE) has taken a stab at the problem, last month releasing a brief Best Practice Guidelines. In brief, the guidelines recommend lower resolutions for feasibility studies, and higher resolutions for construction documents.

3"-resolution (coarse) laser scan

Different stages of a project gain different levels of accuracy:

Level Purpose         Precision     Accuracy       

D     Feasibility     Undefined     n/a

C     Planning        3"            0.25'

B     Design          1/2"          0.10'

A     Documentation   1/4"          0.04' (half-inch)

1/4"-resolution (fine) laser scan

The Guidelines note that surveyors should be aware of insufficient control and target checks, vibration during scanning, wind, and shadows from blocked areas.

Download the PDF from http://files.engineers.org/file/MALSCE-Point-Cloud-Standards-V1-9.pdf

 

 

A.I. for classifying point clouds

Learning how to paint by numbers

 

Does AI have to actually incorporate artificial intelligence to be called "A.I.?" With failures of AI  becoming increasingly publicized, some industry commentators have taken it calling it "actually ignorant."

Never mind the insult. Any time I get a press release landing on my desk touting A.I., I cast a gimlet eye upon the text. Most recently, this was from a Berlin firm named Pointly, which, in fact, is owned by a larger Germany company Supper & Supper (as in dinner).

Pointly says that we can upload a 3D point cloud to its servers and then tag items with color, to identify them. For instance, tag walls of houses on a street in blue, roofs in red (the traditional color in Germany for roofs), and the street itself in light blue. I tried it out, and found that I was working with what felt like the flood tool of paint program from the 1990s. See figure below. It did not feel like AI. Or, if there was AI, my brain was the intelligence behind the artificial.

Flood-filling areas of a 3D point cloud

 

So I asked the firm what was going on. It turns out that I had tried it out too soon:

Pointly, which is live now, is a tool for accelerated manual classification of point clouds. Features for automated processes in Pointly will follow and are being developed [currently].

So, Pointly eventually will figure out what I am doing and, hopefully, start doing the color classification itself. "We want to implement self-learning algorithms," the spokeswoman told me, "So that the more point clouds you label the more the neural network learns and the following labeling work becomes way faster, because the network learned from the work you’ve done before."

Pointly, as part of data science firm Supper & Supper, is trying a few different things, of which flood-fill is one. Other examples include A.I.-supported automated object recognition and CAD model generation from laser scans of freeways. Use-cases include the following:

• Tree inventory (see figure below)
• Road sign and lamp post inventory
• Road mapping
• Building modeling with roof & wall surface analysis
• Change detection in gravel pits

Identifying stumps and trees

For more on Pointly, see https://pointly.ai/

 

Autodesk joins Open Design Alliance

IFC, not DWG

 

Autodesk and the Open Design Alliance are today announcing that the former joining the latter for the ODA's expertise in IFC (industrial foundation classes), used for data exchange in the AEC world. 

The news comes as a shock to those who know the history of Autodesk's law suits against the Open Design Alliance, as well as the history of the ODA being formed originally to open up the DWG format, which, at the time, Autodesk refused to document. ('We have an open format, it's called DXF', was the famous excuse from Autodesk at the time.) Given that Autodesk repeatedly for years also sued a number of the larger members of the ODA, I wonder how well the news is being taken internally -- on both sides.

 

IFC Only

Autodesk's only interest in the ODA is for its IFC libraries, now that BuildingSmart has made the ODA the official supplier of APIs and SDKs. BuildingSmart is the organization that defines the content of IFCs.

The move is steeped deeply in irony, as it was Autodesk itself that came up with IFCs (and object enablers) to solve the problem created by AutoCAD Release 13's user-definable objects (a.k.a. custom objects) -- how to exchange data between AutoCADs that don't understand specific custom objects. Autodesk later handed the troublesome IFCs to an external organization, who, over the decades, expanded IFC to become the de facto exchange format among BIM and AEC programs.

To this day, custom objects continue to dog DWG editors, just as IFC remains an unsatisfactory exchange medium, except that there is nothing that improves on it.

 

The Reason(s)

Autodesk's joining of ODA is part mea culpa and part cost savings. As Autodesk said in a blog post today:

In one of my recent posts responding to some of the constructive criticism we’ve received from our architecture customers, we promised to do a better job of listening to our customers, engage in an open dialogue, and do better where we’ve fallen short.

One of the areas highlighted was progress with our products on international data exchange standards, specifically the need to better support IFC, an open file format increasingly used by our AEC customers.

The cost savings come from the ODA charging relatively modest fees to access its APIs and SDKs as compared with other tools providers. Autodesk, I believe, was already using someone else's IFC toolkit, which I suspect may have cost the company more in license fees.

 

No DWG

Another surprise is that the agreement is for IFCs only, no access to DWG or contributions to the same. As it is Autodesk that defines the DWG format, I would have thought they could -- finally -- start contributing to the ODA, so that there is no longer that six-month lag between Autodesk changing the format and the ODA releasing its APIs to  members.

On the other hand, Autodesk is showing less and less interest in AutoCAD. At one time, Autodesk was changing the DWG format every year. Upon complaints from customers and third-party developers, it promised to change it only once every three years. Now the company can go five years and longer with no changes. This is good, as it brings stability to the entire DWG editor industry.

So, perhaps the ODA doesn't need Autodesk's contributions to DWG, especially as it is the ODA has taken the lead in enhancing the DWG format.

 

 

Here's what's new in DWG from Open Design Alliance

New for 2020

 

The Open Design Alliance has taken over in setting the pace for advances in the DWG file format, now that Autodesk seemingly has lost interest in its best-selling software.

At last week's online DevCon developers conference for members of the ODA, we got shovelfuls of information on what's new in DWG. It overwhelmed us journalists in attendance and so we asked for a summary list, which was promised for this week. Here it is.

 

PDF & DGN Import and Export

Improvements to importing PDF files into DWG databases:

• Import full lines of PDF text as MTEXT
• Import linetypes
• Support for clipped regions during import

PDF export now uses graphics caching to improve the performance in cases where multiple views of a model are exported to the same PDF file.

Microstation DGN export from a DWG database is new, so that the ODA now supports both DGN-to-DWG conversion and DWG-to-DGN conversion.

 

Revision Control

ODA’s native revision control for DWG is refactored for easier integration into real-world client applications.

The new implementation allows clients to more easily implement revision history for groups of related files, including non-DWG files.

 

Point Clouds

Conversion of point cloud data to Autodesk's RCS ReCap Source format.

Automatic level of detail selection for RCS point cloud visualization supports fast visualization and manipulation of large RCS point clouds.

 

Model Documentation

New VIEWSECTION, VIEWBASE, and VIEWPROJ commands create model documentation (2D) views of models. See figure.

Documentation views generated from a 3D model

 

Saving model documentation views to standard DWG files is in progress.

Precision hidden-line removal is in progress.

 

3D Modeling

Boolean operations (such as union and intersection) are now supported by the ODA Solid Modeler on OdDb3dSolids in DWG database.

Visualization for OdDbGeoData objects, which includes map data from third-party providers, is now integrated with the ODA visualization framework.

- - -

Contact www.opendesign.com  for more information.

 

How Programming CAD Goes On, Behind the Scenes

One hierarchy, many algorithms

 

From Ralph: We have no appreciation for the hard work that goes into modern life, whether food that appears magically in grocery stores, the tight-knit hardware-software of smartphones, and the code that goes into today's CAD programs. In this blog entry, which first appeared on the LEDAS blog, we get an inside glance of some of the thinking that goes on behind the code that goes into our favorite software.

- - -

by Ivan Rykov, LEDAS Group

Geometrical modeling projects often contain some form of central hierarchy of objects. A geometric kernel, for example, has a hierarchy of geometric entities. Another example is a geometric solver which has a hierarchy of objects and constraints. Simplified inheritance diagrams for these hierarchies are shown in the figures below.

Many routines/algorithms of other modules need to process all the entities in a hierarchy in a similar fashion, although with different implementation details. So, polymorphism naturally is used with these kinds of hierarchies.

Assuming we are talking about C++ (and this is usually the programming language of choice for algorithmic libraries in CAD), using virtual functions is the first consideration for implementing polymorphic behaviour.

However, a variety of routines applied to the entire hierarchy, or even part of it, is usually quite big. When speaking of geometric solvers, this would include translate/update (to systems of equations and back), validation (for constraints), auxiliary object calculations, technical stuff like serialization/deserialization, and UI representation.

A simplified hierarchy of entities in geometric kernel

 

A simplified hierarchy of entities in a geometric solver

 

If we were to place all of these function as virtual functions in the base, each of the derivative classes would clutter the hierarchy, create huge classes and break the “single responsibility” principle.

class Entity
{
public:
    virtual bool Validate() = 0;
    virtual void Update(SystemContext*) = 0;
    virtual void Translate(SystemContext*) = 0;
    virtual void Serialize(Stream*) = 0;
    virtual void Deserialize(Stream*) = 0;
    virtual void Draw(DrawContext*) = 0;
    // ...
};

Besides the problem of loading the base object with too many responsibilities (and thus making an ugly, hard to read interface), there are other problems with such an approach.

One of the problems is that adding a new algorithm to the system requires modifying all of the hierarchy, by adding a new virtual method to every class of the hierarchy. This not only affects compilation time, but also prevents any algorithms from being put into separate libraries/executables. For example, the Draw() method would be better implemented in the GUI application, instead of the core computational library.

Another problem arises when we need the algorithm to handle only the part of the hierarchy (e.g., only the constraints, but not the objects). In this case we either have to place the virtual function in the very base class (but then have an empty implementation in all the irrelevant objects) or else put it into the first suitable successor in the hierarchy. The former option additionally worsens the interface of the base class with irrelevant functions, while the latter requires to downcast objects to the required class in hierarchy, which breaks the whole idea of using the virtual functions.

The problems I mention here were really bothering us in our software projects, so applying a Visitor pattern was very beneficial in refactoring, which allowed us to make good progress in code organization.

From “switch” statement to virtual functions

Polymorphism provided by virtual functions is a core basis of object-oriented design, which allows client code to work uniformly with all objects in the hierarchy, using the interface of the base class. It is considered a major step forward, as compared to switch statements with downcasting.

Let’s look at these approaches with a simple example. (For simplicity, only two classes of hierarchy are shown and all names are simplified, keeping in mind that such names are highly not recommended for industrial code!) We assume a very common scenario, in which the client code works with an array of objects in the hierarchy, holding them as pointers to the base class. The switch statement implementation looks like this:

void SerializeEntity(Stream* s, Entity* e)
{
    switch (e->GetType()) {
    case tPoint: {
        Point* p = static_cast<Point*>(e);
        // put data of p into s
        break;
    }
    case tLine: {
        Line* l = static_cast<Line*>(e);
        // put data of l into s
        break;
    }
    // ...
    default:
        break;
    }
}

/// CLIENT CODE

void Serialize(Stream* s, const vector<Entity*>& ents)
{
    for (auto ent : ents) {
        SerializeEntity(s, ent);
    }
}

The virtual functions implementation distributes logic between classes of hierarchy, but no casting is required:

class Point : public Object
{
public:
    virtual void Serialize(Stream* s) { /* put data of p into s */ }
};

class Line : public Object
{
public:
    virtual void Serialize(Stream* s) { /* put data of l into s */ }
};


/// CLIENT CODE

void Serialize(Stream* s, const vector<Entity*>& ents)
{
    for (auto ent : ents) {
        ent->Serialize(s);
    }
}

It is worth noting that the switch statement version still relies on one virtual function, GetType(). Then we need to downcast to derived classes to extract specific information, such as the origin/direction of lines. A similar approach without using virtual functions is to employ a series of dynamic_casts with checks for non-zero results.

Everyone knows that using casts usually signals bad design, so there is no doubt that the first implementation is not well written. Casts (static) can cause undefined behaviour or (dynamic) require additional checks and slow down performance. Moreover, introducing changes to the hierarchy (such as adding a new type of object) requires programmers to look through all of the code in a search of “switch” and “if-else” blocks, and then extending them accordingly. This is what polymorphism was intended to resolve.

However, as discussed above, the virtual functions approach also has serious drawbacks. It fills the entity’s public interface with tons of functions completely unrelated to one another. Moreover, it spreads the algorithm’s logic (e.g., serialization logic, modelling logic, and so on) between different classes of hierarchy, and so between different files. This makes it harder to refactor, to extract common parts of the functionality — thus provoking code duplication.

So, we would like to have some kind of valid object-oriented solution that still keeps the code of the same algorithm for different types in the hierarchy close to one another (like in the “switch-case” approach), but avoiding the ugly casts. Visitor pattern provides us with such an opportunity.

… and “back” to Visitor

Let’s see how the visitor-based implementation looks like. (Note: in the real code, either Accept() function of hierarchy classes or Visitor class methods have to be implemented in a .cpp file to allow forward declaration and avoid cyclic dependencies; the code shown here is simplified for compactness.)

class Visitor;
class Entity {
public:
    virtual void Accept(Visitor& v) {v.VisitEntity(this); }
};
class Point : public Object {
public:
    virtual void Accept(Visitor& v) { v.VisitPoint(this); }
};
class Line : public Object {
public:
    virtual void Accept(Visitor& v) { v.VisitLine(this); }
};

class Visitor {
public:
    virtual void VisitEntity(Entity* e) {}
    virtual void VisitObject(Object* o) { VisitEntity(o); }
    virtual void VisitPoint(Point* p) { VisitObject(p); }
    virtual void VisitLine(Line* l) { VisitObject(l); }
};

class Serializer : public Visitor {
public:
    Serializer(Stream* s, Entity* e) :_s(s) { e->Accept(*this); }
    virtual void VisitPoint(Point* p) { /* put data of p into _s */ }
    virtual void VisitLine(Line* l) { /* put data of p into _s */ }
private:
    Stream* _s;
};

/// CLIENT CODE
void Serialize(Stream* s, const vector<Entity*>& ents)
{
    for (auto ent : ents) {
        Serializer(s, ent);
    }
}

Instead of a separate virtual function for each algorithm in the Entity class, we again have a just one virtual function, Accept(), responsible to dispatch any algorithm class, derived from Visitor, into the proper subclass of the hierarchy.

All the disadvantages of virtual functions in the base class are now eliminated:

• Entity interface is now clean, containing only one “technical” function for all of the algorithms
• Algorithm class now again holds together all the code related to the algorithm, so code duplication can be easily eliminated
• Visitor subclasses do not necessarily belong to the same module, so, for example, a “Drawer” visitor can now be implemented in a GUI executable directly, interacting with the corresponding canvas/painting classes

One of the drawbacks of the “switch statement approach” is back again: when extending the hierarchy with the new subtype, we need to check all the visitors to see which of them has to be extended with corresponding Visit… functions. However, searching for subclasses of Visitor is a much more regular task compared to looking for switch statements or casts to subclasses of the hierarchy — especially when we remember that some developers prefer to use C-style casts to using C++ casts. In fact, modern IDEs assist in this task, providing the ability to find all subclasses of a specific class.

In any case, in the projects we work with, the main hierarchy is usually formed at the start of the project, while various processing algorithms are added during the whole project lifecycle. Thus, we usually rather add new visitors than extend them with new processing functions. This really is the scenario for which Visitor is best suited.

Base Visitor class: abstract, or with empty implementations?

When implementing a Visitor pattern it is popular to make the base Visitor class abstract, leaving VisitSomething() functions purely virtual. This might seem a smart choice, because it will guarantee that all Visitors will be reviewed when a new class is added to the hierarchy. At the very least, an empty implementation will have to be added to each of the derived Visitors to compile.

However, we strongly prefer the implementation listed above: all functions of the base Visitor class have implementations, calling the Visit of the direct predecessor in the hierarchy. (The topmost class’ function is empty.) This leads to two benefits, as described next.

Implementation for the needed classes only

Some algorithms make sense for only part of the hierarchy. For example, we might want to draw only objects and not constraints. In this case we just implement the corresponding functions for the visitor, and so do not have to tediously list all of the other functions corresponding to the whole hierarchy.

class Drawer: public Visitor {
public:
    Drawer(Canvas* c) :_c(c) { }
    // Draw what you want only
    virtual void VisitPoint(Point* p) { /* ... */ }
    virtual void VisitLine(Line* l) { /* ... */ }
    virtual void VisitCircle(Circle* c) { /* ... */ }
private:
    Canvas* _c;
};

Compare this to the virtual base Visitor approach: you need a Visitor for, for example, just lines and points, but you have to list all, say, 30 classes of the hierarchy with empty implementations. There is a great temptation to write a couple of dynamic_casts instead. On the contrary, if you write a class with just two functions, you most likely will prefer that to the casts.

Common implementation for part of the hierarchy

If the processing of a certain algorithm is the same for all subclasses of some intermediate class of the hierarchy (for example, all constraints must be processed equally), then it is sufficient to implement a corresponding function of the visitor — VisitConstraint in our example. The following code demonstrates a trivial example of a common processing filter:

class ConstraintsFilter : public Visitor {
public:
    ConstraintsFilter(const vector<Entity*>& iEnts) {
        for (auto ent : iEnts) {
            ent->Accept(*this);
        }
    }
    vector<Constraint*> operator()() const { return _constraints; }

    void VisitConstraint(Constraint* c) { _constraints.push_back(c); }
private:
    vector<Constraint*> _constraints;
};

/// CLIENT CODE

vector<Constraint*> cst = ConstraintsFilter(ents)();

The sequence of calls is very elegant:

1. When particular constraint object (e.g., object of Distance class) accepts the derived visitor (ConstraintsFilter), VisitDistance() function is called
2. Since this function is not implemented in ConstraintsFilter, its implementations from the base class (Visitor) is called
3. Due to the implementation we showed above, Visitor::VisitDistance looks like this:

   virtual void VisitDistance(Distance* d) { VisitConstraint(d); }
   

4. In turn, VisitConstraint is a virtual function, which does have overload in the ConstraintFilter; thus this overloaded function is called and so the entity is added to _constraints array
5. Non-constraints fall into Visitor::VisitEntity() with empty implementations

Conclusion

Using the Visitor pattern in projects that have some kind of main hierarchy is a major architecture improvement that we like a lot and currently apply to many of our projects. It allows us to separate algorithms from objects, providing algorithm access to the subclass API to which the object really belongs. A typical project might have several dozen visitors, sometimes implementing just one-to-two functions for certain parts of the hierarchy.

Read more:
Fast Debug in Visual C++
How to Hack C++ with Templates and Friends
On C++ code bloat
Which Missing C++ Features Are Needed Most?

 

 

© 2020 LEDAS Ltd. All rights reserved. Reprinted with permission from https://ledas.com/post/864-why-we-use-visitor-pattern-in-c-cad-projects/

 

Design Master's Revit Survey 2019

Declining use

 

Design Master Software each year surveys MEP [mechanical, electrical, plumbing] designers about their use of BIM [building information modeling] software, which in this survey means Revit -- although "bim" was not defined so that respondents could apply their own definition. Since the survey began back in 2010, the annual results are now able to show a decadal trend.

The decadal trend reported by Design Master is this: that the use of BIM has not changed greatly over the last ten years, and respondents appear to be using it less for some purposes, contrary to predictions that BIM usage would, in general, increase.

The survey does not examine why Revit use is declining. Could it be due to firms discovering alternatives as Revit development stagnates and the cost of using it increases?

Now, be forewarned that the survey sample size is just 85 and respondents are only from USA and Canada, which is small against a number like the 118 firms worldwide that signed the "Rage against Revit" letter. The size of the surveyed firms does, however, cover a wide range, from single-person firms to ones with over 200 employees.

Here is a summary of results from Design Master:

BIM is being used less in 2019 than had been predicted by respondents in 2016.

The most popular use is generating construction documents.

In second places are BIM being used for 3D collision detection (somewhat less use over the years), Revit-format deliverables (falling steadily), and automated design/calculations (up over the years).

AutoCAD and Excel are used a lot to supplement BIM software, although they were the only two alternatives quizzed about.

Most firms do not charge more when required to use BIM; the exceptions are when they are required to use 3D collision detection or deliver work in Revit format.

For the full details and charts, see designmaster.biz/blog/2019/12/mep-bim-2019-survey-results

 

Addressing the problems of digital twins

Marketing loving it

 

Guest editorial by David Levin, isicad

 

Entities should not be multiplied without necessity.
     - William of Occam

Everything should be made as simple as possible, but not simpler.
     - Albert Einstein

- - -

It is difficult to get past the article entitled “8 Myths About Digital Twins Exposed — Here’s the Reality” by authoritative author Joe Walsh. Digital Twins (DT) is a fashionable topic today, actively marketed by lots of vendors. So it is a professional duty of industry experts to help the market understand what it really means.

 

What Digital Twin is Not

The first section of Joe's article lists and exposes eight myths about the technology. Some of the myths listed are

• There is a Single DT
• A DT Needs to Cover the Entire Lifecycle of the Physical Twin
• DTs Are Used Only by Engineering.

These I do not dispute. Others, however, need clarification or, within the whole set of anti-myths, seem superfluous.

I agree with Joe that the myths and uncertainties surrounding DT are beneficial to certain vendors who exploit the DT hype to sell traditional products just by adding innovative-sounding labels. In my opinion also, the hype around DT provides additional food for experts and media.

 

Do “Virtual Representation” and “Surrogate” Clarify Digital Twins?

Exposing the myths is helpful, because it explains what digital twins are not. The market, however, still wants to know what a DT is, and so the second half of Joe’s article explains it by giving a “usable standard definition” that came as a surprised to me:

CIMdata definition:

A digital twin is a virtual representation of a physical asset or collection of physical assets (physical twin) that exploits information flow to/from the associated physical asset(s). 

NAFEMS SMSWG (Systems Modeling and Simulation Working Group) definition:

The digital twin is a digital surrogate that is a description of a physical asset, such as products, processes, systems, people and devices, that can be used for various purposes. The digital twin makes use of data and information from the real-world asset and provides feedback to this real-world asset.

These definitions seemed to surprise Oleg Shilovitsky even more than me. Oleg clearly explained this in his must-read blog, Digital Twin Hype. Here are twoparagraphs from him:

Both definitions use vague words explaining that information can be used to get feedback about real objects’ behavior. The most interesting part is that the approach to use the information to analyze physical object behavior used in engineering disciplines probably as long as these disciplines existed.

Digital Twin is a cool name. I hope you agree. And I like this name.

Unfortunately, marketing did their job well and created a glorified definition and campaigns about Digital Twin that very confusing. In most of these presentations, the words Digital Twin can be easily replaced by something that is more thoughtful and pragmatic such as model, information, simulation, etc. Unfortunately, it didn’t happen and marketing keeps creating even more variations of digital twin marketing. I found these processes really damaging since as a result, we have Digital Hype and not Digital Twins.

 

Why Not Define Digital Twin as 'Advanced Modeling'?

I almost agree with Oleg completely. Almost – because I suggest replacing his words “Digital Twin can be easily replaced by something that is more thoughtful and pragmatic such as model, information, simulation, etc.” with “Digital Twin can be replaced by a model which can effectively communicate with the associated physical object.”

To be short, I use “object” to denote systems, assets, processes - everything which can be modeled.

Those who consider “model” as insufficiently concrete as compared with, say, “surrogate” may be thinking of an everyday concept in which a model is something that only superficially resembles an object in the real world. Yes, cloth dolls and plastic airplanes can in everyday life be called models of people and aircraft, but our industry understands models as defined by science and technology. In these domains, a model is always intended and built to accurately reflect the aspects of physical objects needed to study the object, as well as to control and optimize its behavior. The counterpart is constructed with as much precision as needed for particular applications and studies.

A constructive understanding of what a digital model is assumes that it involves a system that consists of the following aspects:

• Virtual objects that are associated with components of the physical object,
• Functions defined over the virtual objects that adequately reflect the properties of the physical object,
• Relations defined between subsets of virtual objects and adequately reflect relations between components in the a physical object.

In other words, a model is not just a virtual (digital) representation of a real object, but it also is by definition a virtual representation, which within is sufficiently isomorphic to an associated real object. Perhaps this interpretation of the model and its isomorphism may seem to some to restrict the idea of a digital twin. In reality, however, the definition is needed for meaningful communication, and communicating entities must have some shared compatibly interpreted subject domain (knowledge base).

I do not find it useful to imagine a digital twin as a black box, about which it is only known that it miraculously supports effective communication with the object in reality. If we were to have in mind such an understanding of a model, it allows us to conclude that if a digital twin is not a model it hardly can be useful anywhere.

A model-based approach to DT is of course not something that was just invented, even if the word “model” is not used explicitly. For example, see my favorite passage from Michael Park's article “Digital Twins Offer Unmatched Insights For Design Engineers”:

The basic notion is that, for every physical product, there is a virtual counterpart that can perfectly mimic the physical attributes and dynamic performance of its physical twin. The virtual twin exists in a simulated environment that can be controlled in very exact ways that cannot be easily duplicated in the real world, such as speeding up time so that years of use can be simulated in a fraction of the time.

These hyper-accurate models and simulations offer engineers and product designers unmatched insights across the entire product development cycle. Still, digital twins are more than just an evolution of digital models, although their goal is similar: Higher quality products and better product support at less cost and less effort. 

I like this clear and pragmatic explanation and do not consider “DT are more than just evolution of digital models” contradicts my emphasis of a model-based interpretation of a DT: it is not just evolution of modeling, it is the next higher level of modeling.

I believe that a digital twin need not a “hyper-accurate" model necessarily, as its usefulness does not directly depends on the number of parameters involved in modeling. I agree though that the hyper-accurate DT of an aircraft is hyper-valuable.

Is DT actually something new? A characteristic feature of DT is its capability to support regular bi-directional communications with associated physical objects. Note that such communication could be implemented by a human who shuttles between a model and the object being researched/controlled with the help of the model. A more pragmatic case is the communication between two people: the operator of a model and the operator of associated physical object. Perhaps such a process can be called a digital twin when, for example, it is the simulation of an (emergency) situation on a spacecraft carried out by the flight control center.

In fact, the fundamental novelty of the digital twin lies in the new level of efficient communications with physical objects (typically through Internet of things as some people believe). The key word is “efficiency” as it reflects the capabilities of today’s technologies that enable online, real-time communications between many objects of real world and their complex models – communications that can be implemented with practically any required level of performance, density, and reliability.

At the end of his article, Joe notes that, based on the clear definitions from CIMdata and the NAFEMS SMSWG, various forms of digital twins and their definitions can now be considered. He concludes:

There are multiple forms of digital twins for a wide variety of purposes. Each form of a digital twin has specific characteristics to meet its intended goals and should have a more detailed definition outlining the specific capability, functionality or implementation associated with that form of digital twin. The ASSESS Initiative has established a key theme and working group around Engineering Simulation Digital Twins and has provided the following definition of an Engineering Simulation Digital Twin:

"An engineering simulation digital twin is a physics-based virtual representation of a physical twin (physical asset or a meaningful aggregation of physical assets) that exploits information flow to/from the associated physical twin."

I do not consider the two basic definitions of DT to be intelligible. I completely agree that each form of DT should have a more detailed definition, and I think that if in the definition of Engineering Simulation Digital Twin we replace “virtual representation” with “model,” everything will be OK.

 

“Digital Twin” As a New, Useful Notion

So, in my point of view Digital Twins is a new advanced level of model.

Why would those involved in marketing DT avoid characterizing it as a model? The reason is clear: effective marketing requires new exciting words, and “model” is something too old, too familiar, and too boring.

Is it useful to denote new form of modeling by a new notion? New terminology is harmful if one considers the orthodoxy in the Occam's Principle quoted by this article's starting epigraph: "One should not multiply things unnecessarily." If, however, marketing honestly explains to the market the real and potential capabilities of new modeling tools, then the new notion can be considered useful.

The concept of "digital twin" is useful, and it will be successful when it stimulates the market. If we do not admit that something essentially new is behind a largely speculative concept, then we probably oversimplify the state of affairs and so do not listen to Einstein’s "Everything should be made as simple as possible, but not simpler" (Princeton books). In effect, when we want to follow Occam's principle, Einstein warns we should not throw out the baby with bathwater -- in other words, we should not go too far in our simplification.

Left: Occam; right: Einstein

A similar hype can be observed around "artificial intelligence." It is impossible to define AI constructively; therefore, we observe monstrous journalistic and marketing speculations abusing it. At the same time, however, useful AI solutions are gradually emerging in a fragmented way. Some of them even demonstrate useful approximations to what human intelligence can recognize as intelligence not produced by humans.

No amount of misleading hype can prevent the market from creating really new solutions and truly useful products that in time gradually will justify names as brassy as artificial intelligence and digital twin.

- - -

[This article is reprinted from isicad.net/articles.php?article_num=21408 with permission.]

Google, Stalker

Cyborgs vs. humans

 

When someone, who you don't want, follows you around, it's called stalking. The sole reason Google does anything these days is to stalk us. From killing software, to merging software, to offering new services, all have just one theme: to wrench more data out of us.

The reason behind killing useful products like Picasa? Google couldn't use Picasa to collect data on us, so it had to be replaced by online Google Photos, where Google could stalk us in real-time.

Google reads the metadata of each photo we store with it online. Metadata is information stored in each picture file, and it includes the obvious (date and time) and the less obvious (location and camera model). From that, its computers "learn" what locations look like, and when you were where. Take pictures of having fun at a restaurant and Google learns more about your travels and how busy the restaurant than it does by everyday tracking through your phone.

Location and unique ID metadata stored in a photograph

Instagram does the same when it urges you to write in a location: when you type in the location information, you are doing the work of fine-tuning a location that Facebook may have already gathered from the camera's geographic metadata.

Instagram asking me to tell Facebook where I was

 

(For a time, Instagram thought I was posting photos from Brazil. I wasn't, but I sure wish it would continue thinking so. OTOH, I am pretty sure that data stalkers like Facebook pose wrong answers to trick you into enter data. For a time, Facebook repeatedly asked me if I had gone to university in Lyon, France. That tactic was all it took for me to no longer fire up Facebook.)

Facial recognition, first launched in Picasa, links disparate photos together, as Google learns what it can from groupings of people. In Picasa, it was touted as a way to quickly find all photos of one person, but Google also had you doing learning work for it (for free) when it asked you, "Is this a picture of Jane? Is this another picture of Jane? Is this also a picture of Jane?"

Picasa sill asking for people's names after facial recognition is turned off

Evil

Google stated it wants to be the world's foremost information provider, but in order to do so it first must be the world's foremost information collector.

Marketing in the company's early years positioned the company as altruistic. "Do no evil" was a clever phrase designed to disarm the population from thinking too hard about what its software was doing behind the scenes.

I first realized Google wasn't being just the best search engine during the Google Books fiasco. Google offered authors and publishers a chance to host our books, and in exchange would pay us revenues from accompanying books sales. I liked the idea and was, in fact, a beta tester for Google Books, posting many of my ebooks even before it went official.

Then a serpent invaded the garden. Google announced that a shocking 30% of each book would be available free to the world. A page or two free, sure, but 30%? It got worse. By searching just four times, someone could read the entire book free. Then, it got worser. There never was any revenue from Google.

My HTML reference on Google Books

Eventually, we writers realized the purpose of Books was to read all the information in them to feed Google's voracious search engine. We see the same tactic occurring today when Google sees no value in paying newspapers to carry their content. Eventually, I pulled my books from Books (except for my HTML references book, for which I have received $0); I wonder if Google keeps copies to peruse at its leisure. When you erase something from the cloud, you can never know for sure whether it's actually been erased from the cloud.

That incident soured me on Google's initial idealism.

For me, the final straw came about with my experience using Google's voice input. Typing text on a phone is annoying, especially with the number of mistakes I make. So I tried voice input. At first, it was pretty bad. But I had read it takes a while for the google to learn your voice. So I gave it a chance, and then I was shocked.

Within two weeks, Google was transcribing my voice perfectly. Far from being pleased with the increased efficiency, I felt horror at the intimate intrusion of "me," as expressed externally by my words. The monster behind the multi-colored logo had gotten to know me too well, and I turned off voice input for good.

 

Resignation

So while I am resigned to using Android, I do all I can to frustrate Google's omniscient data collection. I turn off as many data trackers as I can in Android, being fully aware that Google remotely turns them back on from time to time and through dark UI practices "hides" options (figure below) and keeps settings turned on by duplicating them.

Google, dissatisfied with all it knows about me, wants to know more

 

I use Bing. I use Opera. I use VPN. I don't link devices and accounts. I minimize my use of GMail, Google Drive, and Docs. I don't use their other cloud-based offerings at all, such as Google Photos.

Should it come to it, I've bought a spare pair of non-Android phones that also act as long-range walkie-talkies.

Cascading

I suspect that the two founders of Google never intended to become Stalkers of the World. But, really, they had no choice. Consider the cascade of steps involved:

i. To provide all information, they needed to set up systems that could collect all information.

ii. To ensure the information was pure (unbiased), they had to set up systems that collected information surreptitiously.

iii. To maximize the use of their systems, they had to make their data collection systems pretty much free (Android, Chrome, Search) and then tout benefits while obfuscating drawbacks. As well, free meant they could attain a market share in data collection as close to 100% as possible.

Stalking is illegal. In Silicon Valley's twisted view, however, stalking is the highest good. Like victims who are stalked but receive no outside protection, all we can do is minimize our exposure to the stalker.

 

Revenge

I close with that curious incident which occurred shortly after Google got its first grown-up CEO. He was to have an interview with a publication, and so naturally the reporter did the right thing by researching his interview subject and learning intimate details using Google's own search engine. When the CEO learned of this "invasion of privacy," he boycotted the publication.

At the time, we scoffed at the elderly CEO's nativity about life in the Brave New World we were championing, but we could not know that the day would come when we would begin to feel like him. I wonder: was his final revenge on us having his company learn all about our intimate details?

 

 

ZwSoft Cuts Prices and Raises Prices

Targeting Autodesk

 

ZwSoft is offering free upgrades to network licenses. In doing this, it is aiming directly at Autodesk's discontinuance of our industry's normal approach to network licensing. (Autodesk now only sells a license per person.) The Chinese software company has software for DWG editing, machine design, some verticals, and for portable devices. It offers permanent licenses, maintenance subscriptions, and rent-to-own licenses.

 

How Network Licenses Work

Network licenses allow anyone at a firm to run as much CAD software as they need. Some might need CAD for just a few minutes a day, others use it all day. With network licensing, people who don't need it very much are not penalized by paying a full price, as Autodesk has now imposed.

A network license typically specifies the maximum number of instances of software that can run at the same time. If the limit is 10, then ten people can operate the CAD program at the same time. Network licenses are particularly useful for firms that employ shift work: ten users during the day, ten more during the evening shift, and ten during the graveyard shift. 

Because many more people can use the software than can with stand-alone licenses, network licenses tend to cost 30-40% more than stand-alone ones. Under Autodesk's new scheme, the firm would pay 200% more by having to purchase 30 licenses.

To track of the number of copies of software being used, firms deploy network licensing software on a central server. At times, there might only be one or zero people making use of the ten licenses. Maybe ten is too many -- or too few. Network analysis software logs the demand for software so that you can adjust the number of licenses appropriately.

Companies like Graebert Gmbh check your license use at the end of each month (of Kudo, in their case), and then accordingly adjust the amount you pay. In their scenario, you don't pay for too many licenses or work with too few.

Knowing the Unknowns

The ZwSoft offer, which runs until the end of the year, offers network licenses at no extra cost. If you license 20 or more copies, then they throw in a Network License Analysis Tool, normally $500 extra.

The fine print, however, notes that "The license is time-limited,"

[Sept 2, 2020: ZwSoft wrote me with some additions and corrections.]

so we don't know how long the network licenses run for free. At some point, ZwSoft might send you a bill.

On its Web site, ZwSoft publishes no network license prices, so we have no idea what the savings are. We know, however, that stand-alone prices are going up. because...

"On our Network License offer, it is a free trial through 12/31/20. At that point, users will have the option to purchase. We will not issue any charges, unless they explicitly purchase. The price to upgrade to network licensing for STD is $225 and $50."

Prices Up

A few days before announcing free network licenses, the company announced price increases with the release of ZwCAD 2021 on September 1. In general, prices go up $200:

Perpetual licenses

• ZWCAD STD from $799 to $999 $899
• ZWCAD PRO from $999 to $1199

Subscribe-to-Own (after 4 months)

• ZWCAD STD $200/month to $250/mo $225/mo
• ZWCAD PRO $250/mo to $300/mo

The company says it has 900,000 users.

 

The Target

With Autodesk making stumbles like emphasizing named-licensing, failing to port all its software to the cloud, and de-emphasizing development of legacy software like Revit, competitors are lovin' it. Earlier this year, Bricsys, for example, released a video educating users on the drawbacks to named licensing. During a media event last month, Siemens emphasized that design software does not belong on the cloud. Over the summer, nearly a hundred architectural design firms worldwide signed a letter publicizing the problems they face using Revit in 2020.

Being closer to Wall Street than to Main Street is not the way to keep customers content.

 

My First Software Manual -- 1981

Job Control Language

 

Cleaning out my cupboards last week, I came across something unexpected. Software documentation that apparently I wrote in 1981. I have no recollection of it, nor do I know why I wrote it. All I recognized was the name of the co-author, Linda, a classmate of mine during that, the last of my seven years at the University of British Columbia.

The document is historical not only for it being the first software documentation I ever wrote, but also in how it was produced.

1976

My first direct experience with computers was UBC's mainframe, an Amdahl that was plug-compatible with the then-standard IBM/360. In a huge linoleum-floored, fluorescent-lit, noisy cardpunch machine-filled room, we undergrads wrote our brief programs into punchcards, one line of code per card. A deck represented the entire program. We ran the deck through a card reader, and then waited on the result. With dozens (hundreds?) of undergrads running programs on that single shared computer, the wait tended to be longish.

The results of our programs came from an enormous line printer that commanded the center of the room. It was called a "line printer," because it printed an entire line of text at a time -- 132 characters at once on tractor-fed green-white stripped paper. It clattered noisily but was pretty fast. It commanded the room, because we gathered about it as we hoped our programs would not generate results that contained the words of dismay, *WARNING* or *ERROR*.

Whenever the printer encountered a problem, it slowly and magnificently raised its hood at us. Foolish undergrads who, while waiting for their printout, placed their card deck atop the printer saw the cards sailing through the air like a game of 52-pickup. The rest of us waiting for our results erupted in loud cheers at the sight.

1980

There was no interactivity; this was batch computing at its coarsest level. Or, as we called it at the time, "jobs." Grads, however, accessed the mainframe through interactive terminals. I still recall the feeling of awe from the first time I entered the terminal room. It was bathed in very subdued lighting, had carpeting on the floor and individual cubicles housed green-screen terminals with padded office chairs. Hushed, with only the quiet clicking of keyboards. It felt like the engineering deck of an UFO.

Having access to a terminal also meant we could store files; no more card decks flying through the air! Although the first year or so, I had no concept of the difference between data files, programs, the operating system, eh?. I did love the commands I now could deploy: CREATE EARTH (create a file named 'earth'), then KILL EARTH (erase the file). The power!

At first the terminals were limited to line editors. To edit text with this kind of text editor, you would enter the number of a line, use control keys to move the cursor along the line, and then make the change. Pressing Enter committed the change to the file. (Early versions of MS-DOS were no different: the text editor Microsoft wrote was called EdLin, and it also edited text just one line at time.) Clumsy, frustrating, and only slightly better than using punch cards.

1981

During my last year, the University of British Columbia Computing Department experimented with a full-screen editor. This was an amazing advance: now we could move the cursor anywhere on the screen to make an edit. For me, however, as a touch typist, this was a multi-million-dollar computer finally catching up to my $400 typewriter; the difference, of course, being that I no longer needed to retype pages.

So while programs were now stored as files instead of punchcards, the output was done in exactly the same way. We had to wait for a printer to spit out results. Less conveniently, the printer was in the basement of the computing center, while our lush terminal room was two stories up. At least we got the exercise of running up and down the stairs.

The printer was one of the first laser printers in the world. Made by Xerox, the 9700 was as long as an automobile. It offered one font, Courier, and printed at 300dpi. However, it was blazingly fast: two pages a second. As we waited for our output to be pigeon-holed by the computing center staff, we could hear it quietly making that tick-tick sound, twice a second, for each page being printed. When done printing a job, it added a cover sheet identifying the job as ours, and then stapled the pages together.

The computing center developed the full-screen editor further, adding features like bold (*B*) and italics (*I*) codes that we embedded into the text and different fonts. I never did figure out how to specify a different font; Courier was good enough.

And so we come to this piece of documentation, my first. I have no idea why Linda and I were tasked with writing it for the GVRD, the Greater Vancouver Regional District (now Metro Vancouver). At the time I was taking classes in the faculty of Community and Regional Planning, and my engineering specialization was transportation, so perhaps a prof had us do it as a class assignment.

No matter; I loved to write and the full screen editor was the breakthrough I'd been waiting for since I learned to touchtype a decade earlier in Typing 9. Editing without retyping!

The manual described the JCL (job control language) syntax for running a program that simulated land use and traffic flows. Running the program cost 37 cents a second. Running the program from a terminal added a surcharge of 40%. Data was stored on tapes that cost $1 to mount and then 30 cents a minute while mounted. Printouts of the result cost 3 cents a page of paper plus 4.8 cents for 120 lines (a page of text). A typical run cost about $20 (about $50 today, with inflation). Today's cloud computing regresses to those bad old days.

1983

When two years later I got my first computer, a Victor 9000, it came with no software other than what MS-DOS v1.x included -- the EdLin line editor, with which I wrote. A brilliant friend who camped in my apartment for a couple of months while waiting for his job to begin wrote me a full-screen editor in MS-BASIC. (That was all I asked from him in terms of rent.) To make it fast, he wrote it as a single, large array that held all the text. This approach limited files to about ten pages, max, but that was all I needed to be productive. Thanks, Norbert! I added function key definitions that added formatting and so on.

The true brilliance of the personal computer was that it no longer cost each time I needed to run a program. Sure, the computer (and printer and cables and diskettes) came to $6,000 (about $15,000 in today's money), but after that it was free to use.

1985

Little did I realized that four years after I left UBC I'd leave my position as a professional engineer -- forever -- to become a professional technical writer in September, 1985.

I was there to watch the progression from line editor to text editor to word processor to desktop publishing.

 

 

VM on Linux in ChomeOS -- too slow for me

It's not Elementary

 

Linux running in ChromeOS gets a bit more functional with each release of the browser-based operating system. It still starts, however, in terminal mode, although Web sites will tell you how to install a GUI (graphical user interface) so that you don't need to type endlessly-long command-option strings at the prompt, such as this one:

sudo apt install qemu-kvm libvirt-clients libvirt-daemon-system bridge-utils virtinst libvirt-daemon virt-manager -y

Now in 2020 it is possible to install VM (virtual machine) software on ChromeOS's Linux subsystem, and then run multiple Linux GUIs. ChromeUnboxed, for instance, shows how to do this with one Linux package at chromeunboxed.com/forget-windows-i-just-installed-elementary-os-on-my-chromebook-and-its-awesome/. I recommend you read the comments to the article to learn about steps the author forgets to detail.

You do need a Chromebook with a powerful CPU, significant RAM, and more than the usual storage space, as a VM typically wants 20-40GB for itself. My Acer runs an i7 CPU with 8GM RAM and 128GB of storage -- about as powerful as it gets these days.

The writer says, "I want to point out that the Linux operating system runs like a champ inside the virtual machine." I found it didn't. Any click I made took Elementary a minimum of three seconds to respond; sometimes the wait for me was ten seconds. After playing with it for a half-hour, I stopped the virtual machine, and then uninstalled it entirely.

I suspect running Windows on Parallels on ChromeOS will be just as painful. There's just too many layers of code at work.

 

 

 

German station furnishings for AllPlan

DB + BIM

 

If you like trains as much as I do, then you might be excited to learn that Germany's Deutsche Bahn rail operator has released all of its station furniture as BIM files, free. That's 3D models of guard rails, shelters, sign posts, light standards, and so on. The package includes IFC exchange profiles and planning documents.

S-Bahn station in Berlin (photo by me)

 

There numerous catches in the offer from DB Station & Service AG (the arm of Deutsche Bahn that operates the train stations in Germany) and Allplan:

• The registration page, download site, and software install are all in German
• The download is an EXE installer file that searches for AllPlan
• The installer stops when it cannot find AllPlan

Otherwise, cool.

Some stats about the Germany passenger rail system:

• 5,400 stations
• 800 station buildings
• 3,300 tenants, like stores, in 700 of the stations
• 1 million m² (ten million square feet) floor space
• 20 million passengers/day
• Using BIM since 2017 to manage it all

To access the library: info.allplan.com/de/dbahnsteigplanung-in-allplan-db-station-service-ag-projektvorlage

https://www.allplan.com/en/

 

Nanosoft ships 3D DWG editor nanoCAD Pro 20

At about half the price of AutoCAD LT

 

Last week we heard of this year's release of parametric MCAD from TOP Systems in Russia. This week, it is Nanosoft's turn.

Nanosoft has two CAD systems, the lower-priced 2D-mostly DWG editor named nanoCAD Plus; version 20 shipped last month. The other is the higher-priced 2D/3D DWG editor named nanoCAD Pro; version 20 shipped this week.

I am biased towards it, because I've creating tutorial videos for the company, and so I know the software rather well. It offers a remarkable price-performance ration: for about half the price of AutoCAD LT, you get:

• Parametric 3D modeler with history
• Direct modeler and editor
• 2D and 3D constraints
• Import and display 3D point clouds and IFC files
• Import PDF files as underlays or as entities
• Uses DWG as its native file format

This for US$865, which includes three years maintenance. nanoCAD uses DWG technology from the Open Design Alliance, and the 3D modeling/parametrics kernel from C3D Labs.

For this year's release, here are some of the new and improved functions in nanoCAD Pro 20:

New:

3D constraints -- creates assemblies from parts

Direct modeling -- pushes and pulls faces to interactively alter models

Dynamic UCS -- automatically reorients the UCS for drawing on 3D faces

Drawing comparison -- shows differences between similar-looking drawings

Bounding prism -- clips 3D models for seeing inside buildings and mechanical designs

Script editor -- indents and color coding for editing and running LISP, VBA/ActiveX, DCL (dialog boxes), and JavaScript routines

Improved:

Dynamic input -- enters command options right at the cursor

Undo/Redo -- previews changes made by the Undo and Redo commands

For an illustrated list of new and changed features, see the What’s New page at nanocad.com/products/pro/updates/. 

Download a 30-day trial version from nanocad.com/products/pro/download/

 

 

Report: Autodesk ignored Revit concerns in 2014

Australia, New Zealand

 

Last week's list of Revit complaints by an international group of architectural firms reads much the same as ones listed already back in 2014. As reported by Iain Godwin of Godwin Consulting, "In 2014, a group of 12 large Australian and New Zealand Practices sent a letter to Autodesk’s Sales Director, Autodesk Asia Pacific, stating,"

We feel that the value for the money that we have been spending on the software subscriptions has been diminishing steadily, most noticeably in the last 2 or 3 years.

We also feel that the annual enhancements to the software are not being sufficiently driven by existing user requirements.

It appears that marketing driven changes are much more dominant.

According to Mr Godwin, "The net response from Autodesk was, no letter, no email, no phone call. Nothing." 

 

In and Since 2014

Tim Waldock, Revit consultant at RevitCat in Sydney, Australia writes on Linked-In (linkedin.com/pulse/autodesk-letter-tim-waldock/) regarding the "Rage against Revit" open letter to CEO AndreW Anagnost, noting Autodesk's lack of action in 2014, and then goes on to list the good and the bad changes to Revit since 2014. He concludes:

It will be interesting to see how Autodesk actually substantiate their claim: "Engaging, listening to and addressing the concerns of our customers is a top priority for Autodesk."

The Linked-In post includes the 2014 letter in its entirety.

ITC releases IntelliCAD 10.0 to members

by Randall Newton, CONSILIA VEKTOR

 

The IntelliCAD Technology Consortium (ITC) announced the release of IntelliCAD 10.0, which it says is a “major update” with many new features. 

The ITC creates a core CAD product that other vendors use to create custom drafting software. ITC products support the DWG file format popularized by Autodesk AutoCAD. As a member of the Open Design Alliance, ITC uses ODA code libraries to assemble parts of its software.

David Lorenzo, ITC President, stated, “IntelliCAD 10 is a huge structural update with new core libraries from the ODA, a new geometry library, updated graphics, and new APIs. Yes, there are several interesting new features for users too, but the structural changes provide a much stronger foundation on which ITC members can build even more complex and diverse applications.”

New In Ten Point Oh

The OpenGL ES graphics device is now supported for enhanced display quality and performance, which also includes new visual styles. A new materials library contains materials that can be assigned to entities and layers and then displayed with visual styles; spot and point lights can also now display in-program with visual styles. The redesigned View Manager has more options available and allows users to make selections dynamically for the view’s background, perspective, lens length, etc. 

New dynamic input incorporates command prompts and measurement guides at the cursor.

Printing with .pc3 customization files is enhanced with the ability to set margins, page size, .pmp files, and settings that vary according to the output file type.

New import file formats include .obj and .stl.

Many new properties are available for various entity types in the Properties pane, including the ability to view properties of custom entities.

IntelliCAD 10 also includes the ability to migrate user interface customizations from previous versions.

 

Tweaks Here & There

Smaller features that will have a big efficiency boost for end-users are

• the ability to re-order and sort layer filters
• combine paragraphs and clear formatting of multiline text
• determine which user interface items are hidden when cleaning the screen
• use new videos and online resources from the Start page
• convert closed entities to solids
• specify a layer state for each published sheet
• a redesigned status bar
• enhancements for Section 508 usability
• better 4K display support
• an initial Hungarian translation

 

IntelliCAD for BIM

For versions of IntelliCAD that work with BIM files, draw roofs and beams, create mitered joints for beams, explode attached BIM underlays into polyface meshes and polylines that can be edited while retaining underlay functionality, and use new options for section lines and elevation lines.

 

IntelliCAD for MicrosStation Users

For versions of IntelliCAD that open, save, and edit .dgn files in their native format, the supported command set now includes Multiline Text, Chamfer, and Dimension Styles.

For advanced users, IntelliCAD 10 incorporates ODA SDKs version 2020 Update 2 and enhanced .NET and IRX functionality for increased API capabilities. Other improvements and fixes are found in performance, working with files, editing and selecting entities, text, printing, and more.

 

More Goodies

An overview video of IntelliCAD 10.0 features is on YouTube at youtu.be/FmyaEugUMio.

IntelliCAD 10.0 is supported on Windows 10 through 7, 64-bit versions only.

IntelliCAD Mobile Viewer versions are available for Windows desktop, Google® Android, Apple® macOS® and iOS.

For more information about IntelliCAD and the ITC, including case studies, please visit www.intellicad.org/.

[This article reprinted with permission from https://www.consiliavektor.com/2020/07/31/structural-updates-highlight-intellicad-10-0/]

Expect Creo-on-the-cloud in a few years

The shoulders of Atlas grow longer

 

When PTC purchased Onshape, it wasn't just to help Onshape get rid of its huge debt to early investors. That was just one reason.

As PTC ceo James Hepplemann put it last week, "If the Onshape acquisition seemed expensive, I mean, it was." It was, after all, not just the $470 million that's been talked about in public; PTC also paid off the remaining debt on top of that, perhaps as much as a further $169 million. The $470 million was just the net purchase price.

Onshape boosts PTC's efforts in boosting recurring revenue, sweet talk for more subscriptions from more customers. As Mr Hepplemann explains, SaaS [software as a service] "is a long-term monetization effort that’s significant, because while we’ve brought our customers from perpetual to subscription, there’s another and potentially bigger monetization associated with bringing them from on-prem [software running on customers premises] into SaaS [software instead running on servers rented by PTC]."

More SaaS = more revenue for PTC = more cost to customers.

Onshape boosts PTC efforts to do more SaaS. The technology on which Onshape runs becomes the technology on which PTC's other software will run. PTC has named this technology "Atlas," the Greek's mythical god whose punishment was to shoulder the entire firmament for eternity. He was one unhappy camper. (Image source allposters.com.)

"We’re proceeding full speed ahead on the Atlas program, which aims to generalize the underlying Onshape SaaS architecture and put it to work more broadly across the entire PTC product portfolio," explained Mr Hepplemann.

PTC says its Vuforia AR software already runs on Atlas; no surprise, as that was an easy port. As does Onshape, of course.

 

Creo of the Clouds

Mr Hepplemann calls Atlas "a breakthrough CAD platform that will allow Creo [MCAD] and Windchill [PLM] to get into that SaaS world years earlier than they would have independently, and on a consistent platform. Everything’s knitted together on a common platform. But you know, I want to be clear: it’s not a short-term strategy."

The interesting claim that PTC plans to port Creo to Atlas is interesting because it is pretty much now a given, in 2020, that CAD belongs on the desktop. Nevertheless, Mr Hepplemann is giving his company several years to pull cloudCreo out of the hat.

Don't expect it this year or next or the next. "Let me be clear," repeated Mr Hepplemann. "It’s kind of a mid- to long-term thing, because we’re really looking at doing kind of a major redevelopment of those properties onto this platform."

Here is what he is promising customers: "The key thing for us would be to optimize [cloudCreo] for compatibility and continuity" with desktop Creo. Also:

• A product that looks a lot like Creo
• Runs in the cloud
• Is multi-tenant
• You open up the same files
• You work with the same user interface

 

Creo vs Onshape

The question then becomes, How does MCAD package Creo compete with MCAD package Onshape, when there is no differentiation with both running on the cloud? From earlier comments by PTC, it appears that Onshape will always have fewer capabilities than Creo, be cheaper than Creo, and be targeted at smaller firms than Creo. The little step-brother.

Mr Hepplemann employs a Google-Microsoft analogy, in which Onshape is Google Docs and Creo is Microsoft Office:

I have an analogy I’ve shared internally many times. If you think of Google Docs, it’s a completely fresh thought, fresh look at an office productivity suite. Meanwhile, you had Microsoft Office, which was on-prem [desktop].

But Microsoft came out with Office 365, which was SaaS, [because] companies like PTC might have been thinking about whether we should go to Google Docs. Many do, and particularly new companies, but companies who have been using Office forever really do prioritize compatibility of data and user experience, but we like to [also] have SaaS.

So you could imagine we could build a product that looks a lot like Creo. But it runs in the cloud and is multi-tenant, but you open up your same old files and you go right back to work with the same old user interface, much like you did with Office 365.

Meanwhile, Onshape over here is more like Google Docs, unbridled, unconstrained innovation, no legacy, trying to invent the next generation of application. So I think Onshape fights the fight against competitors, and Atlas allows Creo and Windchill to bring their customer base [to the cloud].

Future of Onshape

Onshape is not compatible with Creo, although Atlas is supposed to make sure of that. So, I wonder if Onshape has a future at PTC? In my version of alternative history, once PTC gets Creo to a 'good enough' level on the cloud, the lower-revenue-producing Onshape might be, I wonder, seen as a drain on sales and marketing resources, and so dumped from the line-up. 

 

[Quotations with some editing for clarity and brevity sourced from https://seekingalpha.com/article/4362125-ptc-inc-ptc-ceo-jim-heppelmann-on-q3-2020-results-earnings-call-transcript]

 

 

 

 

 

 

 

 

 

Letter to Anagnost Gains Country Representation

17+18+10=45

 

One of the complaints directed at the group who posted this week's open letter to Autodesk CEO Andrew Anagnost was that the list of 17 architectural firms were primarily British. It could be positioned as a regional complaint. The complaint involves an up-to-70% increase in subscription fees (over five years) for Revit, insufficiently developing the software to handle today's architectural projects, and other issues.

With the publicity of the letter, more architectural firms learned of the so-called "Revolt against Revit" and so signed up to also support the list of complaints against Autodesk. The following countries are represented by architectural firms:

• USA
• South Africa
• France
• Canada
• Netherlands
• Norway
• Israel
• Australia
• Great Britain

The original letter was hosted by 17 firms; 18 more signed the letter this week as a result of the attendant publicity; ten firms agree with the letter's content but say they fear retribution from Autodesk if their names were to be made public.

The letter has been posted online as a downloadable PDF file: https://letters-to-autodesk.com/

To learn more about why 45 firms are expressing their discontent with Autodesk publicly, be sure to read Martyn Day's 4,000-word report on the conditions leading up to this week's event at https://www.aecmag.com/comment-mainmenu-36/2046-autodesk-aec-customers-demand-better-value.

 

Europeans Examine Non-European BIM Vendors' Sales Practices

Autodesk having its own 2020?

 

The European Construction Industry Federation (ECIF) is a go-between organization of construction firms and 32 countries in Europe. It is calling on the European Commission to investigate the non-competitive nature of BIM software in Europe.

Specifically, the organization is calling out software companies involved with BIM that

• Are non-European
• Offer software only by increasingly-expensive subscriptions
• Limit software use to only in the country of the software's licensing
• Sell software with poor backward compatibility
• Have little actual experience with construction

When we take these criteria into account, I am of the opinion that it seems that the ECIF is referencing just software vendor specifically (despite the plural) that I think might match the criteria: Autodesk. It could be that there are others, as well, but Autodesk would be included if there were a group coming under the spotlight.

Here is what ECIF wants, and it is something we all would want:

It should be for the market to decide, and this is why the construction industry needs open, competitive, and fair markets for software and related services.

The FIEC's position paper is not new; it came out in February. Still, it seems that Autodesk might be having its own version of 2020, as architects revolt publicly against the limitations of Revit, and the EC  being called to investigate BIM vendors that aren't European. 

ECIF makes more points, which go against globalization (of which I am no fan, and so the Balkanization is interesting to observe). ECIF wants that

• EU companies should be able to have their data hosted on EU territory, by EU servers, under EU legislation
• Creation of a secure European cloud
• Non-EU software services should be required to meet EU standards for interoperability and open access
• European Commission should promote open standards for data, protocols, and file formats in public procurement
• Legal protection of BIM data to protect intellectual property

I dislike government getting involved in our personal lives and company practices. But as we see with the likes of Google and Facebook, the network effect creates authoritarianism of a corporate kind, and that must be fought against. 

You can download the position paper from https://www.fiec-ar.eu/images/pdf/WG4.0-2020-02February-D-FIEC%20position%20paper%20on%20Software%20companies.pdf

 

Top Systems updates T-Flex CAD for 2020

T-FLEX CAD 17

 

Top Systems makes one of the advanced MCAD design system that originate in Russia, and so tends to get little attention outside of the country.

T-FLEX CAD was one of the earliest to adopt parametric design, uses Parasolid as its kernel, and now release 17 adds the following functions:

• Advanced surface and curve modeling to make surfaces of any complexity
• Accelerated complex assembly design to load the aircraft model 8x faster
• New solutions for simulation, such as Inertial Relief 
• Advanced electrical design tools
• Enhanced markup functions
• Import Revit and FBX files

On top of the base MCAD software, Top Systems offers the following add-ons:

• T-FLEX Analysis
• T-FLEX Dynamics
• T-FLEX CAM
• T-FLEX Electrical
• T-FLEX VR (see below)
• T-FLEX Nesting
• T-FLEX DOCs for PLM

The software runs on Windows, and is available in English and other languages. A 122-page PDF tells you what's new in this release: tflex.com/resources/whatsnew/. Unlike less flexible CAD vendors, you can license the software standalone, networked, student, or with annual maintenance.

Details here: www.tflex.com

 

What Statistics Can Tell Us About SARS-CoV-2 Test Results

by Nikolay Snytnikov

 

To better estimate the situation with СOVID-19 within our company LEDAS, we decided to make antibody testing available to our employees, all of whom had been working  remotely since March. Fifty-three of our 70+ employees were interested, and so we organized out-of-lab testing thanks to Invitro’s laboratory service. Both IgM (early antibodies) and IgG (late antibodies) were examined, and the results were simple: all had zero positives for IgM, and one employee was positive for IgG.

We were able to congratulate ourselves for the zero result in early antibodies, meaning we are quite safe. But now we needed to focus on the late antibodies, as they provide more interesting information. Since we at LEDAS are now involved in more medical software projects, we considered this an opportunity to apply statistics to our medical data.

 

Test Accuracy

It is important to keep in mind that the accuracy of SARS-CoV-2 antibody (ELISA) tests is not very high: the sensitivity (true positive rate) of our tests was ~95% and the specificity (true negative rate) was ~98%. These are due to the lack of a perfect verification procedure, because double-checking through PCR (polymerase chain reaction) testing is not precise either. The specificity could be increased more easily by using old blood samples collected before the coronavirus era, such as from 2019, which could be considered non-infected.

Statistics are more accurate when working with big numbers, and so results for groups are more meaningful. By contrast, individual results are not particularly indicative; for individuals, they could be even counter-intuitive. I’ll look at both cases.

 

Individual Results

Let’s assume that the general level of infection in the population is L=5%. This corresponds to results from a nationwide study in Spain published recently in the Lancet journal. Spain is one of the highly infected countries.

When someone gets a positive result, there are two possible reasons that are mutually exclusive:

• TP – there is an infection, and the test is a true positive (accurate)
• FP – there is no infection, but the test is a false positive (inaccurate)

The probabilities P(TP) and P(FP) are calculated in this way:

P(TP) = level of infection * sensitivity = 0.05 * 0.95 = 0.0475
P(FP) = (1 – level of infection) * (1 – specificity) = 0.95 * 0.02 = 0.019

The probability of getting a positive test P (Positive) is the sum of the two probabilities:

P(Positive) = P (TP) + P (FP) = 0.0475 + 0.019 = 0.0665

This means that if you are in Spain (or in another country with the same level of infection), and you applied for a test of this type, then you have a 6.65% chance of getting a positive result.

What can we conclude from this? That we should apply the conditional probability formula to measure your chances:

P(TP / Positive) = P(TP ∩ Positive) / P(Positive) = P(TP) / P(Positive)

The probability that you are actually infected in the case where you received a positive test will be approximately

P(TP / Positive) = 0.0475 / 0.0665 = 0.714…

So, before taking a test in Spain, you have a 5% probability of having been infected; after testing, despite getting a positive result, you still have only a ~71.4% probability of being actually infected. Your chance of being not infected after a positive result is ~28.6% – not so small!

So what is the benefit of testing for the virus? To understand this, we should consider how the probability of being infected changes when the result is negative. Once again, there are two possible reasons for testing negative:

• FN – a patient is infected, but the test was false negative
• TN – a patient is not infected, and the test was true negative

The probabilities are:

P(FN) = level of infection * (1 – sensitivity) = 0.05 * 0.05 = 0.0025
P(TN) = (1 – level of infection) * specificity = 0.95 * 0.98 = 0.931

The probability of getting a negative test is the sum of the two:

P(Negative) = 0.0025 + 0.931 = 0.9335

This conforms with the 0.0665 probability of getting a positive test result, as P (Positive) + P (Negative) = 1.

The probability that you are actually infected when you test negative is

P(FN / Negative) = P(FN ∩ Negative) / P(Negative) = P(FN) / P(Negative) = 0.0025 / 0.9335 = 0.0026…

So, while before test you had 5% of probability of having been infected, after getting the results you have either

• 71.4% of probability of being infected, if the result is positive, or
• 0.26% of probability of being infected, if the result is negative

In the one case, you increase the probability by an order of magnitude; in the other one you decrease it by an order of magnitude. The difference is significant!

It is easy to see that if the specificity of the test somehow decreases to less than 95%, then even after getting a positive result you will have less than a 50% chance of being truly infected, which is rather counter-intuitive.

 

Group Results
The results for our test group can be described using a binomial distribution. Let’s input our parameters P (Positive) and N=53 and see what WolframAlpha tells about this:

What is the probability that there were no infected employees in our company, given one positive? For this to be true, 53 test conditions should be satisfied simultaneously and independently:

• One condition (the only positive result) is actually a false positive; the probability of this is P (FP / Positive)
• 52 conditions (the negative results) of which each is actually a true negative; the probability of this is P (TN / Negative)

The product of these 53 Bayesian probabilities is

So we have a not-so-small 25% chance that no one was infected.

What is the probability that many employees were infected? This is not as easy to calculate, but it is easy to make an upper-bound estimate. To have more than N-infected employees, we should have at least N false negative tests, based the one unit of difference between them provided by the single positive result in all our antibody testing.

The probability of having N false negative tests is not more than

If N=2, it is already less than 1%, so it is highly unlikely that we have more than two infected employees. Using this formula, the probability of having exactly two infected employees is less than 14.2%.

Therefore, the highest probability is that there is only one infected employee at LEDAS, as we have a 60+% chance for this. This implies that there was no transmission of the virus inside our company.

 

Some Notes About Assumptions
I did the computations with these assumptions:

• The statistical selection was done independently
• The percentage of people in the population that contracted coronavirus is 5%

The first assumption is, of course, not fully correct, because our employees had been working from home for a long time, and so have a low probability of being exposed to the virus. So, the selection was not independent.

The second assumption of 5% infection does not correspond to the official data for Novosibirsk region, in which our staff is located. As of mid-July, there were seven thousand cases among 2.8 million people, for an infection rate of about 0.25%. On the other hand, it is well-known that official testing data don’t include asymptomatic and false negative cases; this is true for all countries, and so Russia and Novosibirsk are not exceptions here.

We can adjust the infection rate without doing extensive antibody testing, but instead using estimates from the infection fatality rate (IFR). Some studies suggest that it should be within the range of 0.5-1%. The official number of coronavirus-caused deaths in the Novosibirsk region is 142 (as of July 12). We can calculate the number of all COVID-19 cases in the range of 14-28 thousand.

Unfortunately, these numbers look underestimated, as the official number of death cases might not be correct. For example, some deadly pneumonia cases are “false negatives” or are not considered to have been caused by COVID-19 as the primary cause of death.

Better estimates of lethal COVID-19 cases can be calculated by comparing the mortality of this year with that of the previous year, as was done by the Financial Times for certain countries and regions. The raw data is available on Github.

For the Novosibirsk region, the data (available elsewhere) suggests there were an additional 400 deaths in June 2020 (when compared with the monthly average for 2019). Of course, not all deaths this year can be attributed to COVID-19, but very probably it is the primary reason of the increase. Assuming that 300-400 deaths were due to COVID-19, then the number of all cases can be roughly estimated as 30-80 thousand per 2.8 million of people. This is in the range 1%-3%, and much higher than official data of 0.25%.

 

Estimates from Dunbar's Number

The numbers can be double-checked with a very different consideration: sociology and anthropology. The so-called Dunbar’s Number proposes that people can comfortably maintain about 150 relationships with other people.

Dunbar’s Number can be useful in understanding the level of disease spread in the population: if you know of no one among your friends or relatives who contracted the disease, then the likely number of cases is less than 1%. But if there is at least one person you know personally, then the rate is higher than 1%.

In our case, it looks like the 1-3% estimate is correct, as for many of us we know of two to three COVID-19 cases among friends and relatives. Very probably during the next few weeks and months, some of our employees will contract the disease as well, due to the increasing number of infected people in the general population.

Below you can see how the probabilities of different cases for one person change if we decrease the parameter “infection level” from 5% to 2% (midpoint of the 1-3% interval). It is interesting that for the positive result you have nearly 50:50% chances of true positive and false positive:

Conclusion

We believe that there were no transmissions inside our company, because we implemented remote work in March, well before there were any cases reported in our city. Restrictions from authorities began more than ten days later. And we are taking care to not contribute to transmission of the virus in the future.

Stay healthy, safe, and take care!

- - -

Nikolay Snytnikov is the chief operating officer at LEDAS. He supervises projects in the areas of 3D cloud applications, medical software, CAD, and BIM, and the development of the LEDAS Cloud Platform. http://ledas.com/en/

 

 

RetroCAD is Hip

The DOS-based reawakening

 

Hipsters like old stuff, grungy stuff. Wrecked boards, worn out mugs. Best present I ever got for one of my daughters, according to her, was bringing back a slightly-damaged-looking camping mug from New Zealand. A present to end all presents. No doubt there is somewhere in China a factory with people whose job it is to make brand-new mugs look banged-up. Or maybe the look is part of the mould.

Never mind. Retro is hip, even in CAD. We have people restarting computers from the 1980s and 1990s running CAD software from those days. If you grok Nth Engine, then you're in the know.

 

RetroCAD Is a Thing

The RetroCAD YouTube channel, for instance, has 40 videos about getting software like AutoCAD 1.4 (my first version in 1985!) and KeyCAD running on old hardware like a Compaq 486 and Kurta tablet -- all under DOS, of course. https://www.youtube.com/c/RetroCAD/videos

I have a old '386 sitting under my desk, waiting for me to have time to tinker with it. It has the all-important 5.25" and 3.5" diskette drives. And my bookshelves groan under a large collection of 1980s-1990s software that encompasses everything from Quarterdesk Desqview to an early VersaCAD. The #1 killer in getting old software resurrected is whether the magnetic patterns on diskettes survive the decades.

Remember inserting 15 diskettes to install one program?

 

Generic CADD Renaissance

People keep buying the ebook version of my "Learn Generic CADD 6.0" book, even though it is 27 years old. That's because there is a renaissance in running the beloved CAD program on today's Windows and old time DOS systems. A purchaser from last week told me, "Just fired up my old copy of generic cadd to do some drafting and realized my memory is not as good as I thought it was in using it. This [ebook] will come in handy to refresh my brain cells."

You can get tips for doing this from https://worldcadaccess.typepad.com/blog/2013/01/running-generic-cadd-in-2013.html and its 46 comments. The ebook (PDF) is $14.50 from https://www.worldcadaccess.com/ebooksonline/2015/07/gcp.html

About Nth Engine

Nth Engine was an expensive, memory-laden accessory board that sped up AutoCAD by storing the display-list in RAM. As their ad copy put it, if you have to wait more than two seconds, then you're waiting too long. Nowadays, redraws and regens tend to be instant, so two seconds seems today w-a-y too long to wait. But not in the 1980s. I timed the famous nozzle.dwg drawing taking 4.5 minutes to regenerate on an 8088 (early PC); 30 seconds with the optional, extra-cost math co-processor. So, yah, two seconds was a breakthrough.

At the time, I was still technical editor  at CADalyst, and I got to review the Nth Engine board. I found it wasn't as fast as some of its competitors, and so the magazine lost the full-page ads as a result. By the end of the 1980s, hardware display-list processing was superseded by software ones, and then by the early 1990s Autodesk killed the thriving industry when it hired the main programmer away from Vermont Microsystems, got sued, paid $6 million in damages to VMI, and killed the industry anyways.

Remember bird's-eye view? Ya, that was part of that era as well.

Graphics used to be horrendously expensive. Think $1000 - $3000 per component for "high" resolution of 1024x768. That's a thousand or three for the graphics board, a couple thousand for the 19" monitor. And sync was fixed, so if you wanted the change one of them, you had to change both. A good, fast, large plotter was $15,000. Now multiply those prices by 2.5 to account for inflation today.

The CAD software was priced reasonably; the hardware was a killer. Today, the roles are reversed.

BNC connectors, anyone? How about configuring a serial port?

 

 

The Big Release from C3D Labs for 2020

C3D Toolkit 2020

 

In this era when it appears some CAD vendors really just want your money in exchange for not much, it's good to see annual releases that burst with improvements.

How bad is the just-give-us-money scam? (And it really is a scam.) At one CAD event I watched this year, the 2020 version of the software had exactly one new function -- and the marketing team didn't talk about it. [Audience cue: groan.] It almost seemed like the pre-event chatter at the firm consisted of "Since our software's update is so lame, let's tell reporters about anything but."

Others, I am pleased to report, are bursting with goodness and newness. Solid Edge 2021 from Siemens is one; C3D Toolkit 2020 from C3D Labs is another. 

'Course, the software from C3D Labs is totally out of your league. You and I wouldn't buy it, ever, 'cause it's meant for programmers. The "Toolkit" part of the name is the clue. If you want to develop your own CAD system or other software that works with CAD data, then you'd reach for a toolkit, like the ones from C3D Labs, Open Design Alliance, Siemens, Spatial, and others. C3D Labs says their toolkit, however, is the most complete one on the market, with these modules, and you can license each module separately, should you need the services of just one or two:

• Geometry kernel -- C3D Modeler
• Constraint manager -- C3D Solver
• Polygonal mesh to B-rep converter -- C3D B-Shaper
• Visualization engine -- C3D Vision
• Data exchange -- C3D Converter

Surface Modeling

The big addition to this year's release is surface modeling. Surface modeling is used for industrial design, when you want swoopy curves. Think car bodies. One of the new functions generates a variety of surfaces based on sections and control functions, as the figure illustrates:

1. Conic section curves
2. Sections in the form of NURBS
3. Sections in the form of arbitrary splines
4. Circular sections
5. Ruled sections

We get a hint of where C3D Labs is thinking of heading with surface modeling in future releases, with this sentence from its press release: "This is considered the high-end of CAD functionality and paves the way for solving design tasks in the aerospace and shipbuilding industries." So they seem to be aiming for the lofty levels currently handled by the Dassaults and Siemens of the world.

And why not: once you have all the math in place, geometric kernels are done. Really, the only new problem that Dassault (through Spatial) and Siemens (through Parasolid) are tackling these days are 3D models that combine solids and meshes, and how to edit them as if the two representations were the same thing. Progress is, on that front, slow, giving chances for competitors to catch up. I note that C3D Labs has a 'polygonal mesh to B-rep converter', which in some future release could well become a 'polygonal mesh to B-rep editor'. It is, after all, just math.

The other changes to this year's release of C3D Toolkit are in the areas of solid modeling, sheet metal, direct 2D editing, 3D visualization, interoperability (file translation), and the aforementioned mesh to b-rep conversion.

For instance, developers can now add a function that punches sheet metal with any arbitrary shape, and the punched portion can have a different thickness than the original sheet metal.

For the details, consult https://c3dlabs.com/en/blog/products/c3d-labs-brings-advanced-surfacing-to-c3d-toolkit-2020-update-targets-high-end-3d-engineering-softwa/

 

Updated: Lenovo announces first workstation with 64-core CPU

Lenovo ThinkStation P620

 

Summer used to be quiet time for user conferences, press events, and product launches. People were, after all, on vacation. Fall was the busiest time for me, but spring was busy for others who cover tech events. Then a thing called coronavirus happened, and people said that "afterwards everything would change."

For these events, it has changed. Last week and this week, I received invitations for five online events -- from Graphisoft (2 events), 3D Systems, Siemens, and Lenovo. In July! That I am travelling and on vacation during summer months -- well, that's not a factor. I have to miss three of them, precisely because I am travelling (annual visit to mother-in-law), but some events have replay enabled, fortunately. 

One live event I am missing is Lenovo's launch of what it calls the world's first workstation with  64 cores, thanks to AMD's new Ryzen Threadripper PRO CPU. For me this is a "finally" moment, as AMD of late has been blowing Intel out of the water. AMD always was a CPU maker and I seem to recall some law suit that forced Intel to share 8086 code with AMD so that its CPUs are compatible with Windows, et al. I've watched Intel since the late 1980s and have been amazed at its ability to stumble, most recently its inability to get its CPU size down to 10 manometers; competitors like AMD and Samsung are already down to 7 and 5nm. 

Update 1: Here are the specs for the three Ryzen CPU models announced this week (image source The Register):

So, yah, AMD is blowing Intel out of the water, according to what the tech publications say, in both advances and pricing. Lenovo is pleased to ride the wave by being the first to offer this new ThinkStation P620 workstation with an AMD CPU that offers:

• 64 cores
• clock speed up to 4.0GHz
• 128 lanes of PCIe 4.0 bandwidth
• 8-channel memory support

Outside of the new workstation (image source Lenovo)

In addition, Lenovo stuffs the box with

• 10GB built-in Ethernet
• support for up to two NVIDIA Quadro RTX 8000 or four RTX 4000 graphics cards
• up to 1TB RAM (or 512GB, depending on which Lenovo document you consult)
• 20TB of storage
• PCIe Gen 4 workstation

Back of new workstation 

So, this is the most cores ever offered in a workstation, according to Lenovo. This kind of a computer is meant for software that actually makes use of many cores, such as rendering and FEA [finite element analysis]. In addition, the software has to support 64 cores, as some are written for fewer. It is not too useful for CAD drawing and editing, which is strictly single-core work. Some CAD systems use more more than one core for a very few, very specific tasks, like loading drawing files and regeneration (which is reloading drawing files).

Inside of new workstation

This new computer ships in September, with a price that is unknown to me today. The liberal use of the words "up to" means that the base price will be reasonable (for a workstation), and then skyrockets. More info from http://www.lenovo.com/p620

Update 2: The ThinkStation P620 starting price is US$4,599 (USD) and includes:

• AMD Threadripper Pro W3945W (12/24-core model)
• 16GB RAM
• 256GB M.2 PCIe 3x4 SSD
• NVIDIA Quadro P620 GPU
• Keyboard and mouse
• Windows 10 Pro
• 3-year onsite warranty

 

Is saying Carbon Copy even legal anymore?

Not CO2-based

 

We have mainstream media and eager-to-please-greens politicians spouting the need to reduce carbon-this and carbon-that, never realizing that they are talking about the wrong thing. They mean CO2, but out of a plenitude of ignorance shorten "carbon dioxide" to "carbon." One is a molecule mostly found in the gaseous state, the other an element found most often in the solid state. I leave it as an exercise to the reader to determine which is which.

Perhaps the switch in scientific nomenclature is deliberate, because carbon is normally black and dirty, while CO2 is clear and clean. Better to appear to be seen calling for cleaning up that which is black and dirty, than that which is clear and clean.

We are reassured by those in the know that "carbon" is evil, evil, evil, and why hasn't lazy government spent enough on eradicating it yet. Never mind that CO2 is the chemical that allows vegans to remain vegan.

And so it is with that background that an auto blog that I follow made reference to carbon copy:

...focusing instead on building an updated model influenced by the utility vehicle, but not a carbon copy.

Surely some kind of genetic or chemical manipulation must be involved in making a carbon copy, no doubt. I suspect people even in their 40s might not know the meaning of the term.

In the days before computers made trivial the printing many copies of a document, we typed with typewriters. If we needed a copy of a document, such as for a committee meeting, we didn't want to type it over many times. Instead, we used carbon paper -- very thin paper coated with the graphite form of carbon on one side. We placed it between two sheets of paper, and then typed. Presto, two copies for the effort of typing one. Amazing productivity booster.

If we were desperate, we would use three or more sheets of paper, with two or more sheets of carbon paper. But the text would get increasingly fuzzy with each layer of paper, as they spread out the impact from the typewriter heads pounding away.

And so that is how the phrase "make a carbon copy" came to mean making a near-perfect reproduction. No CO2 needed.

 

PS: You can today still purchase carbon paper, but the embarrassing-to-some word 'carbon' has been removed, and so it is renamed "graphite transfer paper." Graphite is carbon, as is diamond. But you knew that.

PPS: Carbon paper would help detectives in 60s tv shows solve crimes, after the bad guy's secretary threw used carbon paper into the trash can, leaving a perfect copy of the incriminating letter for detectives to discover.

PPPS: If you were cheap like me, you would reuse carbon paper until it no longer could make reasonable facsimiles. From facsimiles we got the word 'fax,' which became the next generation of document copying. But then you knew that, too.

PPPPS: Because we couldn't edit letters when we wrote them (without rewriting them, and who wants to do that), the post script (PS) method was devised to add more material to letters to which we already had affixed our signatures. 

 

Q&A with Graphisoft Executives

ArchiCAD 24 launch

Graphisoft has for years held an annual by-invitation-only conference for its top customers, to which we in the CAD media were also invited. With this year's coronavirus canceling in-person events, Graphisoft opened this year's conference to all users.

Two days earlier, Graphisoft held a press conference moderated by Graphisoft senior manager of global communications Julianna Gulden with Graphisoft ceo Huw Roberts, Nemetschek Group chief division officer Victor Várkonyi, and vp for product success Akos Pfemeter.

We got an early heads-up on ArchiCAD 24 and new product logos, along with a extended question and answer session.

New logos for Graphisoft products

 

In brief, this year's release of ArchiCAD offers enhancements in the areas of

• Enhanced MEP modeler
• BIM change tracking
• Greater file support
• New surface materials and residential library
• Python API for automation
• Param-o as an easier version of GDL for driving parametric parts

The launch will be world-wide in all languages and includes updates to BIMcloud server and BIMx viewer, naturally, but also deep integration with Risa (structural design for North America) and Scia (structural analysis and design for Europe). We first heard about this Nemetschek-wide initiative last year, as well as during last week's interview with the Vectorworks ceo.

BIMcloud server is the core of integrated design for ArchiCAD and this year it gets

• Open for files from any source, making it interdisciplinary instead of just limited to ArchiCAD
• Connect to it from other software through a new API
• Performance boost

BIMx model viewer works on mobile devices and Web browsers:

• New 3D visualization engine that handles files of any size
• A single BIMx app with freemium and subscription options
• Free version offers all forms of viewing, smart measure, and presentation gallery
• Paid version adds large model support, and API, ability to save favorite

- - -

Q: Does the Graphisoft-Scia-Risa integration use the new internal file format? Or does it use OpenBIM?

A: The Nemetschek Group is using an internal data scheme but it also is being developed as an open format for data sharing and real-time teamwork.

There are couple of different layers for integrated design:

• Teamwork for connecting people and workflows -- managed by BIMcloud
• Parameter level, in whcih an element in a model is used by different people for different purposes, which cannot go out of sync or be misaligned.
• Schema is the SAF (?) schema, open, works with other software top

This makes it possible to integrate the structural model inside the BIM model, like MEPs checking that ducts don't go through columns, and the architect knowing from the structural analysis that the building won't fall down. They are- working inside Risa and Scia, which are integrated into a single model with ArchiCAD. BIMcloud ties it together

Q: Param-o looks like Grasshopper. Is it based on it, or a different product?

Param-o parameters driving the design of a table

A: They are two different products with two different purposes, but they can be connected. Grasshopper is for algorithmic design; Para-o is an ArchiCAD function inside ArchiCAD 24 to create objects for architectural products. They look similar but have different purposes and have different functions.

Our roadmap will allow Grasshopper algorithms to drive Param-o and so drive objects in ArchiCAD.

A: Architects are very familiar with visual scripting metaphor of Grasshopper. That is where the similarity comes form. At the technology level they are entirely separate things. Param-o creates parametric custom objects without needing GD, which can be difficult to program.

Q: How does open BIMcloud connect to non-Graphisoft CAD format? Does it handle CAD files from outside the Nemetschek Group?

A: [unanswered]

Q: Do GDL objects work with Param-o?

A: [unanswered]

 

Q: Can you talk about owner hand-off and digital twins?

A: Nemetschek plans to cover the lifecyle of the building, so we are getting into operations side of things. It is too early to talk about this.

A: Support for OpenBIM is a great means to anyone in any process. We think that is a real advantage we have in flexibility and processes we can support. When you think of hand-off, there is data handoff, data lockdown, things that are frozen for the record, things that stay live. The publishing from BIMcloud is very flexible for hand-off. It gets better when the Nemetschek portfolio is involved.

 

Q: Firms are developing their conceptual tools. How do these connect with ArchiCAD?

A: As an architect, I get it. Sometimes design driven by codes and compliance, sometimes by social drivers, by geometry drivers -- and so many tools are used as input to the design process.

Large firms have the resources to experiment with technology and that will always be the case. The challenge is making a custom tool that isn't a dead-end, that it integrates into the workflow. We ensure we have good APIs so that custom development can be interfaced with the heart of building design, which is BIM.

A: ArchiCAD can work as a conceptual tool, even though it is seen mainly as BIM for formal design. Its API is wide open for experimentation. Also, we partner with Rhino and SketchUp.

Q: Is Graphisoft using ODA libraries for Revit?

A: Yes, we are using the ODA's libraries.

Q: Can Revit use BIMx without ArchiCAD?

A: To use BIMx you have to go through ArchiCAD, because it keeps track of versioning and workflow.

Q: All firms tend to use other structural tools. How will you move these firms to Nemetschek tools?

A: Scia is best for use in Europe and Risa in North America. All of them work with open workflows, but if an engineer wants to use a different one, the experience won't be as good. It is up to Scia and Risa to offer deals to convince people to switch. It is not about price, it is about workflow.

 

Q: Is the CDE [common data environment] from Bluebeam still in the works, as was spoken about a few years ago?

A: The Bluebeam solution does not have the overarching vision that covers everything. Today we think of CDE as a "common design environment." Different firms have different workflows, and so being able to connect and work with as many of those workflows is in everyone's best interest. Our strategy is a universal approach.

 

Q: For which competitor products are you writing plug-ins?

A: We are first concentrating on Nemetschek Group products. We allow connections to other programs, such as other structural tools. We have the most control with Nemetschek brands, but then connecting with Open BIM with anything, even if it just at the IFC level. Revit will not have this level of granular integration.

A: Other firms are contributing data exchange -- there are solutions that are specific to regions. Structural engineers typically use more than on solution, to make sure there is not a problem in one software.

 

Q: Is the Nemetschek schema open? Is IFC granular enough?

A: The Nemetschek schema is designed to be open for other companies to design connections. The IFC scheme is a very rich schema that can contain any information related to BIM. In theory, IFC is granular enough; in practice, it is useful for connecting software to exchange information but our format is complete for designing with.

A: The same information could be in both schema; they overlap but are different topics. The shift is from a disconnected transaction mode, to a share experience where you work on one part of the model and I work on another part of the same model.

www.graphisoft.com

 

Navisworks vs ODA BimNv SDK

 

 

OpenDesign Alliance's BimNv SDK is a software development kit for dealing with Navisworks files and was developed independently of Autodesk.

 

About Navisworks from Autodesk

Navisworks is Autodesk's universal file format for viewing project files made with multiple software programs, such as AutoCAD, Revit, and Inventor, as well as neutral file formats (like IFC, STL, and PDF) and proprietary ones, such as Pro/Engineer, Solidworks, and MicroStation. (See figure.) For more info on formats, see knowledge.autodesk.com/support/navisworks-products.

 

Navisworks model (source: GrabCAD)

For instance, you are designing a building with Revit that contains a machine designed in Solidworks, all on top of topology from a point cloud (Recap). To view how it all fits together, the individual models are loaded and coordinated in Navisworks, from which clash detection, markups, project progress, and so are performed. (See figure.)

 

Navisworks UI (source Autodesk)

The native file formats of Navisworks are as follows:

NWD - contains all the models in a single, compressed file
NWF - contains links to model files, plus markup data
NWC - cache files created when an NWD file is opened, to speed up subsequent loading

When an NWD file is opened, it uses external file readers to view the different file formats; you can substitute your own readers, as the ones provided by Autodesk do not necessarily support all entities. The Autodesk Developer Network offers an API to members ($1,400 and up) for writing add-ons to Navisworks, as well as exporting data out of Navisworks.

 

ODA's Navisworks SDK

The ODA has its own version of a Navisworks SDK that it has named BimNv. The SDK [software development kit] allows programmers to write software that works with Navisworks files. An SDK includes APIs [application programming interfaces], which programs use to make calls that read and process data in the Navisworks files. The ODA says its API is similar to that of Autodesk's Navisworks API. The SDK is available only to Sustaining ODA members or higher ($5,000 and up).

Neil Peterson, the president of ODA, tells me that he sees three kinds of organizations that might prefer using BimNv over Autodesk's SDK:

• Those wanting to view and import Navisworks models in their own desktop software without having to depend on Autodesk software
• Those wanting to visualize and access Navisworks models on the Web without having to use a third-party service like Autodesk's Forge
• Those wanting to export Navisworks models from their applications with no dependencies on Autodesk software.

For example, some government agencies in Australia are mandating the use of Navisworks for certain civil projects, and so the ODA has members looking to BimNv for integrating an Navisworks export function into their products.

The ODA held a Webinar this week on the progress is is making with BimNv API, explaining that it really is a work-in-progress.

BimNv supports NWD, NWF, and NWC files from 2013 and newer. Older formats (2009-2012) are also supported but not as well, because, as they explained, "It is hard to find many Navisworks files from 2009 for testing purposes." Currently, the focus of ODA programmers had been on reading the format, and they have completed some initial editing functions of NWF files.

Navisworks contains data from many programs, so if ODA has already documented one of them, like DWG, then it is OK; if not, then it is more difficult for the ODA to support the data. So, designs made with DWG (AutoCAD and workalikes), DGN (Microstation), DXF, RVT (Revit), and IFC have "excellent" support.

To view Navisworks-style files, ODA offers its Visualizer Viewer (see figure). To view the files, ODA converts them to its own VSF format (Visualize Stream Format). The viewer handles the following functions (among others):

User interface of ODA's viewer (source: ODA)

• Cutting planes
• Multiple models in one viewer
• Save to DWG (quite large, as DWG has to represent everything)
• Export to PDF (when the ODA's Publish SDK is added) and OBJ

Autodesk has no Navisworks viewer for mobile devices, although third parties have viewers that export NWD files to their own viewer's format. So it turns out that the ODA's OpenCloud is the only way to view Navisworks files on mobile devices, through cloud.opendesign.com (see figure).

Mobile viewer from ODA (source: ODA)

What's Next for BimNv

The ODA's programmers are working on exporting VSF files to Navisworks format, reading the clash elements that are stored in Navisworks files, and support for sheets.

 

Q&A

Q: Are PDF files exported from Navisworks files in 2D or 3D?
A: Both, as they is generated by the Visualize SDK.

Q: Has Autodesk a Navisworks viewer for mobile?
A: No. Navisworks files tend to be large as they combine many formats, and so there may be issues loading them on mobile devices. With Autodesk, there are third-party plug-ins to convert the format and then view the files on a mobile devices.

Q: Does BimNv have any accessible BIM data in it?
A: Yes, we can contain data in the file as BIM data, such as those from Revit converted to Navisworks format. Here, some Revit properties are retained, such as colors, categories, and so we can access that data.

Q: Can it visualize without the Visualize SDK?
A: You would have to create your own viewer, such as with your own MSC or Qt application.

Q: Is rendering single-threaded?
A: Yes.

Q: Any plans for multi-threading?
A: Yes, but the priority is not high right now. Visualize SDK could handle multi-threading external to Nv but some output need may to be internal to BimNv.

Q: Is the loading single-threaded?
A: We have both, single- or multi-thread, so you can switch.

Q: Is a trial version of BimNv SDK available?
A: It is available in trial format for Windows only, which is sufficient for concept.

https://www.opendesign.com/products/bimnv

Three, no Four! OSes for Chromebooks

Is Chromebook becoming the ultimate computing device?

The question become relevant as Apple switches it Macs from Intel CPUs to its platform of the future, ARM, and so no more Boot Camp, no more Windows booting natively on Mac desktop computers and laptops. Until now, this was possible, because Windows and MacOS are both coded for Intel CPUs.

The death knell for Windows on new Macs is about two years away. Apple has shown Linux working on ARM CPUs through virtualization; MacOS is based on Unix. It could be the same for Windows, but running an operating system on a virtual machine slows it down significantly. 

Microsoft has an ARM version of Windows, but indicated that so far it is not interested in cooperating with Apple. Apple, likewise, shows no interest in reciprocating. Bifurcation may be the future of computing. https://www.theregister.com/2020/06/25/apple_gives_boot_camp_the/

Apple has shown iOS and iPadOS apps running on Macs, as both would be hosted by ARM CPU.

 

Four OSes, One Computer

Beside native ChomeOS, Chromebooks smoothly handle Android apps, have taken a stab at running Linux programs, and now Google says the lithe computers might in the future run Windows programs, too.

It may seem more complicated getting Android to run on Chromebook than what Apple plans to do with iOS on Macs. Normally, Androids run on ARM CPUs, while Chromebooks on Intel CPUs. However, Google has an Intel version of Android, which allows Android emulators to run on Windows PCs. There even are some dual-boot tablets out of China that run Android or Windows 10.

While I find my i5-based Acer Spin 13 Chromebook a wonderful machine, its lack of running Windows programs can be a practical pain. I took it as my sole computer to a software conference in Europe last year. When the software vendor released a beta of the next release of its CAD system, I could not download it and try it out; I had to wait until I was back home again.

(A workaround is for Mac and ChomeOS computers to use remote-access software, such as TeamViewer or Chrome Remote Desktop, to view Windows programs running on a Windows computer. But this, too, is not a speedy solution.) 

Details about Windows on ChromeOS are, at this point, ultra vague. Given how slowly Linux support is developing, I'd not be one to be holding my breath. androidpolice.com/2020/06/16/google-and-parallels-announce-plans-to-bring-windows-apps-to-chromebooks/

 

Siemens Media Conference: Q&A

 #SiemensMAC2020

 

The first day of the online Siemens Digital Industries Software Media and Analyst Conference 2020 ended with a lengthy Q&A session. All the questions I posted got answered, so that was cool. (That doesn't always happen.) Here are my notes on the questions that were asked:

- - -

Q: Siemens had a sketching app on Android a few years ago, which I notice hasn't been updated in over a year. Is NX Sketch based on this technology?

A: Good catch on that. We learned from CatchBook how people sketch, but NX Sketch is brand-new and is built into NX. It avoids over-constraints on sketches.

Q: Is NX Share available standalone or only in NX?

A: Right now, it is part of NX. It will be available standalone later this year.

Q: Regarding the $100 million AAR [average recurring revenue] of Mendix: why did revenues skyrocket?

A: The Covid crisis is making companies wanting to build apps quickly, to work remotely. Even before, Mendix provided local capabilities such as to Teamcenter. It grew 155% over last year.

Q: What is the pricing for Teamcenter X?

A: We have very competitive and attractive pricing with what's out there on the market.

Q: Is Teamcenter Share a rebadge of Solid Edge Portal?

A: Absolutely not. It is built on the Mendix platform. It adds CAD knowledge to a Dropbox-like platform.

Q: Will Teamcenter be replaced by Teamcenter X over time? Is this part of a roadmap to replace all desktop programs with cloud apps?

A: We are not one of those vendors who go to their customers who say, 'as of May 25 we will be completely SaaS'. This is not our approach. We are modernizing products by bringing them to the cloud. Customers can make the transition, if they want to, when they are ready.

We will never have to have that hard discussion with customers about them having to make a difficult transition of technologies.

Q: What market does Teamcenter X target?

A: Immediately, the mid-market. But there is interest from large customers as well.

Q: How is Mendix affecting development at Siemens?

A: We learn a lot from each other. We don't forget that Mendix began with business applications, such as HR and finance. So we can solve the complete solution across the board for customers [including non-CAD ones].

Q: How is your partnership with Bentley on CALM [Capital Asset Lifecycle Management] going?

A: It is growing business for data management of plant design, offshore platforms, and so on. We will be doing more together, as you will hear in the coming 12-18 months. We are bringing Teamcenter product configuration to these large projects. [See https://blogs.sw.siemens.com/teamcenter/introducing-teamcenter-for-capital-asset-lifecycle-management-calm/ for an introduction to CALM.]

Q: When will authoring [CAD] software like NX be on the cloud?

A: The cloud is a major trend and we are looking at how to bring more of it to our customers. We are working on the authoring side, but the real strength of the cloud is on the collaboration side. So, for instance, we are bringing more collaboration to the authoring software.

Q: What were the revenues of your division for the year?

A: Siemens does not report divisional revenues [such as for Siemens Digital Industries Software], but maybe that will change in the future, like in two years. We are outgrowing our competition every quarter.

Q: How is simulation, CD-Adapco doing?

A: Overall it is doing well. Simulation is core to the digital twin of vehicles and plants. Our simulation revenues are equal to the revenues of the market leader. That gets lost in our large portfolio of software. [ANSYS had $1.5 billion in revenues for fiscal year 2019.]

Q: How many attendees today?

A: Close to 300, our largest ever for a media and analysts day, made possible because of being online.

 

 

Siemens Media Conference, Day 1

 #SiemensMAC2020

 

The day that Siemens Digital Industries Software sets aside for we CAD media and analysts is this year online, instead of in-person in Las Vegas. To make it easier to digest, the conference this year is two hours online on each of two days -- a schedule that I greatly appreciate. 

The Conference interface in the Web browser

The two days seem mostly about how customers use Siemens software, along with overviews of the Siemens approach to the manufacturing industry. President and CEO Tony Hemmelgarn noted that it is surprising how many his competitors force customers into subscriptions. Siemens, he says, lets customers choose their own path, whether how to implement software or how pay for it. Except: I think that the new X-named cloud-based software is available only through subscription payments.

The primary slide is this one, which is an overview of the approach that Siemens takes towards customers:

Siemens Digital Industries Software's offer to customers

The CAD software from Siemens has one advantage that no competitor can offer: software like NX and Solid Edge are owned by a manufacturing company, giving Siemens an edge in knowing what is needed by customers, because it is its own customer. Siemens in its most recent fiscal year made $89 billion, down 8% from the year before.

Me, I'm mostly interested what's new in software. 

 

New Software Announcements

NX Sketch is new software for 2D sketching -- inside NX -- without pre-defined parameters or relationships. Instead, it infers relationships between tens of thousands of curves, including imported ones. In the figure blow, this means that Sketch figures out that all those small holes in the disc rotor (imported from a DXF file as just lines and arcs) are all the same diameter, and that the arcs are linked to the lines and so applies geometric constraints. 

User interface for NX Sketch

Siemens had came out with similar software originally for Android five years ago called Catchbook, but had not updated it in over a year -- I had even paid for it!, In it, you sketched lines and arcs by hand, and Catchbook turned them into basic entities connected by constraints. I suppose Siemens was working on this Windows version instead.

Also new is Teamcenter X PLM [product lifecycle management] software, the SaaS [software as a service] version of regular Teamcenter. (SaaS means cloud-based software paid for by subscriptions.) The new X version of Teamcenter uses modules that customers choose, such as working with BOMs [bills of material], ECM [engineering change management], and release management.

Interface for Teamcenter X

Both are part of the Xcerlator portfolio, Siemens' new (as of last year) name for its collection of CAD software, that includes stalwarts like NX and Solid Edge MCAD, Mentor ECAD, and Polarion ALM [application lifecyle management].

The following slide explains cloud-based Polarion X -- be clear that ALM is totally out of my league.

What ALM entails

Siemens bases its cloud offerings on it Mendix platform (which runs on Amazon cloud or on private servers), which last year made $100 million on its own. 

The capabilities of Mendix

 

More on Teamcenter X SaaS PLM

• Instant-on PLM, because it is runs on Medix cloud and comes with "out-of-the-box best practices"
• Base modules include document management, visualization, workflow, revisions & releases, and basic BOM
• Expandable and personalized with add-ons
• Fixed price
• 30-day free trial available from [to come]

This chart shows the difference between regular Teamcenter and new Teamcenter X:

Teamcenter vs Teamcenter X

Tomorrow we learn about Teamcenter Share, which looks like it runs on smartphones and tablets.

 

Siemens User Conference

Siemens holds their annual user conference online next week. Realize Live 2020 starts at 8:30am PDT over June 23 and 24, 2020 at https://events.sw.siemens.com/realizelive/teamcenter-plm-conference. Siemens expects ten thousand in attendance online. Hope the ON24 system doesn't buckle!

https://www.sw.siemens.com/

 

 

 

Nobody owns 3dconnexion

Gone in 2011

 

A letter to the editor in last week's upFront.eZine newsletter noted that Logitech's new top-of-the-line mouse is a loser. So, the letter writer instead bought a new 3D and 2D mouse from 3Dconnexion, after reading positive reviews. "I am grateful for Logitech's failure because it forced me to look deeper into my other options and allowed me to give my money to another company that I would not have otherwise considered," he said.

In the back of my brain, I recalled that Logitech had spun off 3Dconnexion, and so I let his statement stand. 

Some readers, however, wrote immediately to explain the reader was still giving his money to Logitech, because it owns 3Dconnexion. Did it? Clearly, my back-of-the-brain notion wasn't sufficient in the area of supporting evidence, so I investigated further.

 

Let's Ask the Horse's Mouth

First place to look: the horse's mouth.

The 3dconnexion Web site's Company Profile page has scrubbed any mention of its owners, instead waxes on about how great their mice are. No mention of Logitech. As well, no recent press release mentions Logitech, which a subsidiary would do in order to bask in reflected glory of a normally well-liked company.

 

Let's Ask Google

Next place the look: the oracle of the intertubes. Let's ask Google:

Google's answer

 

Clearly, Google thinks Logitech owns 3dconnexion. So why does 3dconnexion not declare the ownership with pride?

The cached version of every Web page Google finds is the one it has in its storage banks. This is the location from which Google shows you its search results and the information it displays. 

Accessing the cached version of the Web page

 

When I click on the cached version of the page, Google shows me the result from when it trawled the site six days ago. Problem is, the page in its cache from six days ago is the same one I saw today. It does not say who owns 3Dconnexion.

Complete reproduction of Company Profile from which Google should be getting its information

 

So Google is reading one thing, and telling us another. Chalk that one up for the ongoing investigation into the behemoth.

 

Let's Ask Bing

Actually, I use Bing all the time, because I don't care to give Google more information than my Android phone already does. Also, Bing pays me to search with them, Google doesn't. And so every few months I get to present my wife with a $5 Starbucks coupon, and she uses it to buy the biggest latte the chain sells.

So, let's ask Bing.Well, that seems straightforward. Bing does not, however, provide a link to the source for its assertion. It does have a Feedback button, which I clicked and told the microsofties about what I had found (see below).

 

Let's Ask Wikipedia

But what about my notion that Logitech sold 3Dconnexion? Let's ask Wikipedia:

Oops, that page hasn't been updated in ages.

 

Let's Ask Logitech

Let's ask Logitech. Here I credit Henry S, who helped me research this question. He came across the most official reference that there cam be regarding corporate maneuvers: the annual report.

Logitech's annual report for 2011

On page 241, we learn that...

In fiscal year 2011, the Company sold its equity interest in certain 3Dconnexion subsidiaries, and its intellectual property rights related to the manufacture and sale of certain 3Dconnexion products, to a group of third party individuals and certain 3Dconnexion employees.

A full valuation allowance of $2.2 million was provided for $5.7 million of capital loss carry forward from the sale of 3Dconnexion Inc. in the U.S. as the Company determined that it is more likely than not that the Company would not generate adequate capital gains in the next five years before the capital loss expires under the U.S. tax law. The remaining valuation allowance of $0.1 million represents a full valuation allowance for certain foreign tax credit carry forwards in the U.S. 

In short, 3Dconnexion was sold to its employees nine years ago, because Logitech determined it would not be making a profit from it any time soon.

 

I Cancel Twitter

So much like TV 

 

Last week I stopped reading Twitter, cold-turkey.

In the aftermath, most surprising to me was that I did not miss it. I had no curiosity as to what I might be missing. Kind of like when I quit tv cold-turkey at age 18, after realizing I was addicted to it. Not watching tv meant not missing tv.

In the case of Twitter, the urge to quit occurred just as suddenly. I had built up a modest following of 931 (as of yesterday), while I followed twenty or so active users, who  in their commentary were either smart or witty or both. As for me, I wrote about (as a CAD writer) CAD industry press releases, (as a non-fiction reader) philosophical arguments, and (as a media person) documenting hypocrisy in the mainstream media.

Twitter canceled me a few times. It blocked one tweet as "incitement to violence." In the tweet, I had suggested cruise ships could better cater by Millennials by offering bungy jumping out on the ocean.

Twice Twitter stealth-edited me. This is when Twitter shows your tweets in your timeline, but does not display them in the public timeline. I got around the blockage in both cases by posting screen grabs of the offending tweets. And then there were examples from others of Twitter coming to the end of its era in free speech.

One of my followers even demanded that I no longer engage in free speech. Well, that's not quite how he put it. He told me angrily what I shouldn't be writing about. Sorry, but that's not how it works.

For me, Twitter's second stealth-edit was the end. Reading and maintaining my feed took up a fair amount of time, as had tv watching. I did enjoy working out philosophical arguments, because 280 characters forced me to work through arguments very, very concisely. On the other hand, I was receiving no benefit from posting press releases, just as watching tv has no benefit.

So then I thought what if I unfollow most of the twenty, and just keep up with a very, very few uniquely-gifted commentators, like @wretchardthecat. Nope, back to the time sink again.

I still have my @upfrontezine account and will check it from time to time for direct notifications, as well as cross-post from this blog. Otherwise, it's as a girlfriend once explained to me, "It was fun while it lasted, but now it's time to move on."

 

LEDAS Reports on Business Through COVID-19

 

In my nearly 35 years writing about computer-aided design, I realized that CAD is largely immune to recessions. It seemed that each time economic activity declined, CAD use picked up as a way to benefit from the lowered costs. The sort-of recession caused by the rapid spread of coronavirus did a only small dip on the CAD industry.

Here is a report from LEDAS on how coronavirus impacted their business. The company provides software R&D [research and development] services for the engineering market, and released an update on its business in light of the COVID-19 pandemic. The firm is located in a somewhat isolated region of Russia, in the Siberian city of Novosibirsk -- 3,300km from Russia's coronavirus hotspot of Moscow.

Novosibirsk, Siberia

- - -

At the end of March and during April, a few LEDAS customers were forced to reduce costs by suspending or decreasing their projects with us. The decrease was caused by the general global decline, which affected CAD and manufacturing industries everywhere. Personalized digital medicine, a rather new but important sector for LEDAS, also declined, as healthcare changed its focus to fighting the coronavirus, and so other medical treatments were partly suspended.

The decline reversed during May and into the beginning of June from two sources of growth. Some projects returned to their previous levels of execution, and new customers began projects with us. In the area of digital dentistry, for example, one project essentially halted in March but then resumed in May, because dentistry took a big dip during the peak spread of COVID-19 in Europe and North America. Talking of new projects, in May we began a new project in specialized 3D modeling.

The coronavirus-related dip

One reason the impact of coronavirus was rather small on our business is due to our involvement in long-term projects. These projects have long roadmaps, with sales typically lasting years into the future. When you develop a 3D kernel, a constraint solver, or a complex algorithmic engine for polygonal meshes (all in development by our team now), you plan for decades, not for months or even years.

The spread of coronavirus in Russia is not yet in stable decline, but our Novosibirsk region fortunately was not severely affected. Both the infection rate and the lethal rate are considerably less than in the Moscow region and most western countries.

In any case, LEDAS employees began working from home in March. This was due to an internal management decision before the start of the official lockdown by the government. We will continue to work remotely, as our work processes have been adjusted to this new environment.

We consider the pandemic to be the highest level of global threat, and we address it seriously at both the corporate and personal levels.  

www.ledas.com.

[Reprinted with permission from http://ledas.com/news/859-ledas-reports-on-business-through-covid-19/]