Evan on Engineering Software

Ralph Grabowski writes in his blog about UGS supporting Linux with its NX4 CAD release. 

The reason UGS supports NX4 on Linux and X64 not just because customers want it, but also because it *can*.  That is, the NX code base is designed to be portable, and to run on multiple platforms.  It's no big deal (other than writing a check for the time it takes to do the work) for them to support Linux.

Other products, including UGS Solid Edge, Microstation, SolidWorks, and AutoCAD, are designed to run on Windows.  Their developers have written them in such a way that they'd be very difficult to port to a different platform -- whether it be Linux, Mac, BeOS, or Multics (just kidding on those last two.)

It's just a market decision.  I talked to John McEleney, president of SolidWorks, about his company's commitment to Microsoft Windows (and the upcoming Vista.)  He made a good case for it being the right way to create a mainstream product.  While there are ways to build modern CAD products so they are platform independent (I've got some experience in this), if you compare the effort and support costs versus the questionable payoff, it's a hard decision to justify.

Products such as SketchUp, VectorWorks, Cobalt and ArchiCAD are dual platform (Windows and Mac) -- and are greatly respected, not only be users, but also by their peers.  Interestingly, they all share a conceptual modeling tendancy -- which seems to fit the Mac zeitgeist.  But I don't think any of the companies that create these products could justify the work and expense to support Linux, as the incremental payoff might be pretty small. (Opinion only -- if anyone wants to prove me wrong, please do so.)

UGS is in a different position from many CAD vendors -- NX4 and Teamcenter are enterprise applications, not consumer software products.  UGS customers are facing big IT management problems, and big costs in supporting Vista.  For them, Linux is an appealing alternative that will actually reduce their IT costs.  Dassault and PTC are in a similar situation with the enterprise customers for their PLM applications.  You can probably expect to see more Linux support from them in the future as well.

Ralph comments that, perhaps the new definition for high-end (M)CAD "will be (1) it runs on Linux; and (2) it runs on 64-bit CPUs."  I don't think those attributes will define high-end MCAD -- but I think, in practice, high-end MCAD products will tend to have those attributes.

(If you're curious, all the software vendors I've linked to in this post are members of the Open Design Alliance.  If I wanted to cite every Open Design Alliance member that met the criteria I've listed in this post, it would be a very long post indeed.)