Being cross-platform and cross-browser with close-to-native performance is the primary selling point of WebAssembly. What's the point of going that way if you are supporting the exact same platforms you were supporting before? It's also funny how there is no mention of Firefox at all like it's so insignificant we won't even bother testing.

Chrome is the new IE

SketchUp is very cool, but it does not have a constraint solver. If you want to know what that means, try SolveSpace for an easy-to-use CAD program that has one. When you reach its limits switch to FreeCAD which is far more feature complete but harder to learn and use (IMHO).

I found the web version unusable and had to revert to the desktop app. Buggy user interaction was a deal-breaker.

I hate desktop applications. They take a long time to install, they have local file storage that requires me to spend time backing the data up, they eat up disk space on my computer system, they aren't accessible to me if I need to use another computer system, they need to be manually patched and updated, and if I am dependent on them they could restrict my choice of operating system on my desktop.

Web apps aren't perfect either but they have a lot of value to offer users.

There is a large category of users, for whom effortless access through a browser offers great value. For example, in the AEC industry there are lot of roles where one needs to access the 3D model for review purposes and not modeling itself. For these roles a webapp that is quick to open offers actual value (and is not just a cheap gimmick from the vendor).

People say the same things about GUI desktop apps being dumbed-down versions of command line utilities and tool ecologies. Command line tools are still here, and we have more of them compared to the Old Days; it seems focusing entirely on the newest thing provides a skewed view of the totality of what's available, as pre-existing technologies tend to not only not go away entirely, but continue to evolve.

(I wonder if anyone thought command lines were dumbed-down versions of punch-card batch setups.)

3D programs target different domains.

Wings is mainly for creating renderable meshes.

A CAD software's main usage is in modeling physical structures.

The hardware is cheap nowadays compared to the software. Any sub-1k dollar/euro PC will handle most things one would use Autocad for. It's the software that is expensive.

Common codebase for Mac and Windows via an abstraction layer. Same code also transpiles to JavaScript.

Gdpr-shield?

That already kind of happened. A couple former SolidWorks CEOs started OnShape which is a CAD webapp. It's not as mature as SolidWorks which has been around for 20+ years of course, but it's a very capable CAD application.

I believe onshape is developed by ex-solidworks employees.

"export your NURBS to something other"

NURBS does not define a single mathematical entity, like a Bezier curve for example does. You can't convert NURBS trivially from one kernel to another. If the implementation is not exactly the same, the transform is unlikely to be mathematically exactly equivalent.

I think people get into the habit of presenting their software as uniquely capable. And wrapping up their own skills in that particular software. It is similar to a young programmer assuming that Perl (for example) is uniquely capable. In truth talent tends to transcend packages and other factors like ecosystem come in to play.

A talented CAD technician would be useful using pen and a drawing board. We are just not very good at selling those more inate qualities. It is just easier to just sell yourself as an AutoCAD driver.

Of course these sort of packages take a long time to learn, and you are much more productive using something you know. And people adapt their mental model to a particular package. So there is often little obvious point in changing software.

What about Intellicad (https://www.intellicad.org/ )? It seems to be pretty popular among engineers. It reads DWG/DXF (AutoCad format) pretty well. The software itself goes by various names, because Intellicad itself is a consortium. For a nominal fee, you get access to the codebase and can release your own version with your own brand. The thing is shared source, so as a consortium member you need to contribute your changes to the core back to the shared codebase. However this is closed-source.

>> I've never really understood how AutoCAD and Office maintained their dominance.

Saying it's hard to write something that's 100% compatible would be an understatement.

Viewing the same document in Google Apps, Microsoft Word, or Open Office makes me think of QA'ing a website in 3 different browsers.

> I've never really understood how AutoCAD and Office maintained their dominance

Network effects. Once you reach a certain level of dominance in the kind of market they are in, the fact that it easier to find staff that know the product and the people you exchange days with are using the product, the cost of unfamiliarity and potential data exchange errors means anything else has a very high barrier to overcome even to be considered. An incremental improvement isn't enough, you have to either be wordshatteringly better or have near infinite runway and some non-feature advantages to leverage (the latter is how Excel and Word dethroned 1-2-3 and WordPerfect, but who is going to play 1980s Microsoft to dethrone AutoCAD?)

> I've never really understood how AutoCAD and Office maintained their dominance.

Joel Spolsky explained it: most users use only 20% of the app. But each user uses a DIFFERENT 20%. Thus you won't go nowhere until you are almost at feature parity, because everybody will have one little thing you didn't implement, but which is critical for them.

"MicroStation was a hellspawn"

I think you mean is a hellspawn. I think they still have a solid foothold in some sectors.

AutoCAD is nowhere near as dominant as Office is; SolidWorks, CATIA, NX, Pro/E etc all have their users too.

I don't know if it's any cleaner internally but Dassault produces the free (2d focused) DraftSight which is very much on the level of like AutoCAD circa 2000, don't know what the paid version is like or how well it integrates with Solidworks.

"financially viable lightweight enough"

That's the catch, I think. The market is not so large for 3D modeling tools.

The second hurdle is the ridiculous amount of inertia DWG has in the AEC industry. Any solution that would topple Autocad would need to likely provide DWG and/or Revit interoperability to even have a shot at the market.

There are no good options for Revit at the moment so, that's a deep vendor lock in. For DWG there is at least Teigha library from ODA which is based on reverse engineering the DWG format and offers reading and writing of it.

It would be a long shot. But until something changes, Autodesk is just going to keep layering tools, buttons, and menus on top of each other. It's such a rubber band ball now, they would have to start over to make something efficient. But there's no real competition, because even though it's incredibly cludgy, it will work if you learn it's labyrinth.

I've been perfectly happy using Google Sheets for 10+ years even though I'm sure it's no substitute for someone in finance. (I don't think CAD software has the same wide appeal, though)

I know there's plugins and I don't know what the pro version is like but for instance free Sketch Up doesn't even have the ability to draw an arc from three points and while it did have two points+tangent it made some internal and opaque decision on whether it would ever snap to that tangent.

If you're making 2D things, like blueprints for a human to make something, SketchUp isn't as useful. I admit, I haven't tried it in a few years, but when I did before its fundamental purpose was almost polar opposite to my needs.

Well AutoCAD has some tools for drawing 3D but their actual 3D program is called 3D-MAX. This is derived from the old 3D-Studio program that they bought from Kinnetix years ago. Funny thing is in the shop environment 3DMAX is not used. It is the bastard child of Autodesk. Solidworks is predominately used in machine shops these days.They have a SolidCAM that works well for generating G code for CNC usage.

Autodesk also has a newer product called Revit. I have not used it.

Sketchup is intended as a 3D "sketching" tool. The post to go this through in detail would be quite long but the gist of it Sketchup does not scale very far in terms of model complexity. You can model e.g. a factory floor or a complex office building in Autocad while still being able to navigate the model. Not so much in Sketchup.

AutoCAD and Sketchup are very different. AutoCAD is vector based for starters. Zoom into a curve in Sketchup and you find that it is segmented. Solids in Sketchup are made up of triangles. In AutoCAD they can be actual solids.

Wow, $600.

Maybe a reference to AutoLISP?

https://en.wikipedia.org/wiki/AutoLISP