[cad-linux] Re: An open CAD file format

  • From: Eric Wilhelm <ewilhelm@xxxxxxxxxxxxx>
  • To: cad-linux@xxxxxxxxxxxxx
  • Date: Sat, 6 Sep 2003 23:00:06 -0500

> The following was supposedly scribed by
> Massimiliano Mirra
> on Saturday 06 September 2003 06:55 pm:

>xSpace (http://xspace.sf.net), the project I am developing, seems to
>have many goals in common with yours.  Have a look at the page; it's a
>tiny of proof-of-concept, too, but it would be nice to join efforts.
>
>>     http://bugbear.blackfish.org.uk/~bruno/draft/
>>
>> I'd be interested to hear any comments.
>
>I am using dnotify instead of FAM.  This has the disadvantage to only
>work with kernel 2.4.19, but its work is done in kernel space so it's
>probably more efficient.

Both of these are interesting approaches, but I would be concerned about the 
speed of access to large data files in text format vs binary and also the 
precision vs. size trade-off for floating-point numbers.

As Brian pointed out, there will be quite a lot of talk about using a 
networked-database as a cad format in the May '03 archives of this list.

I am a huge proponent or re-inventing (or even eliminating) the smile 
floormat.  I think it is one of the biggest problems in the engineering and 
design world to date.  Each discipline / contractor / player in a large 
engineering project is likely to be using their own flavor of software and a 
lot of documents change hands in proprietary floormats such as dwg or dxf 
(which isn't proprietary, but also isn't any good.)

I've been dealing with some automation tasks mostly in dwg format via the 
openDWG C libraries and a Perl wrapper for these.  Since this project began, 
I have built an object-oriented Perl module for loading, saving, and 
manipulating dwg / dxf files which is currently capable of also saving images 
and post-script files.  I will also soon be adding a database-load/save 
function to this module.  The plan is to use the database to communicate with 
Catia and eventually other software as well.

While I could easily see writing iges / step / xspace / bruno file-format 
functions into this module, I have to ask myself if this is really the path 
that should be taken.

I've gotten a start on a paper attempting to cover this subject:
http://pages.sbcglobal.net/mycroft/proposal/cad_system_proposal.pdf

Apologies for the pdf format and the fact that it isn't complete yet (maybe 
I'll get an update posted soon.)

The thing that keeps coming back to the front of my thinking is how all of 
this is technically feasible, but just needs to be implemented in a way that 
makes it irresistible to users / managers / developers / etc.

--Eric

Other related posts: