As noted in the recent article posted specifically about converting  
geometry [1], a LOT of effort is going into making conversions work  
more robustly.  All of the repairs to date to our existing NMG/BoT  
support are (fortunately) fixes to low-level LIBRT and LIBBN routines  
and have not required changes to our converters.

[1] http://brlcad.org/d/node/82

With the new conversion.sh script, you can actually get a report for  
your own geometry that will tell you which objects succeed and which  
do not.  The usage is very simple and their are built-in instructions  
(conversion.sh help).

We're using that script to fix all of the observed problems with our  
sample geometry as well as running the script on loads of real-world  
target descriptions.  We're presently at a failure rate of about 3%  
on our sample geometry and about 7% on more detailed models.    
Unfortunately, the problems get a lot harder the closer we get to a  
0% failure rate, but that is the goal.  That's literally going to  
take months of effort, so keep an eye out and we'll keep reporting  
progress.

Cheers!
Sean


On Dec 10, 2010, at 8:50 AM, Clifford Yapp wrote:

> It's the latter - we're working on improving the routines, so
> improvements will show up in the existing tools (in fact, right now
> that's the primary way we test them.)
>
> Cliff
>
> On Thu, Dec 9, 2010 at 7:40 PM, Tom Browder <[email protected]>  
> wrote:
>>
>> Have the g-xxx.c conversion program series been updated to use the
>> improved conversion functions, or are the improvements inherent to
>> them because the same but improvrd functions are being used?
>>
>> Thanks,
>>
>> -Tom
>
> ---------------------------------------------------------------------- 
> --------
> _______________________________________________
> BRL-CAD Users mailing list
> [email protected]
> https://lists.sourceforge.net/lists/listinfo/brlcad-users


------------------------------------------------------------------------------
_______________________________________________
BRL-CAD Users mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/brlcad-users

Reply via email to