[Bf-committers] Mesh development freeze

[blender at erwincoumans.com]

Just a bit of context, why I'm pleading for compatibility: 

I'm working on readblend, a library that can extract arbitrary data from 
.blend files. It works great, and I want to rely on this for some of my 
ongoing prototyping and R&D work for Bullet physics. For example, it allows 
me to extract some logic bricks, Bullet soft body data (etc), without having 
to bother about an intermediate fileformat, writing a python exporter etc. 

1) Would it be feasible to add a 'save-as pre 2.50 .blend', using your 
proposed tesselation into the .mface structures? 

2) Can you try to keep the underlying .blend file structure compatible, so I 
can continue my readblend project? 

Thanks a lot,
Erwin 

 

joe writes: 

> On Mon, Apr 13, 2009 at 2:33 PM, Brecht Van Lommel
> <brechtvanlommel at gmail.com> > Can you give an estimate for how long
> this freeze would be needed? 
> 
> My current rough estimate is six to eight months.  People would still
> be free to help out with the bmesh project, of course.  As I
> mentioned, we've been bitten by concurrent work problems before.  The
> "freeze" would end when bmesh is at a state to be merged back in to
> the 2.5 branch. 
> 
> It shouldn't take me more then a few weeks (about equivalent to three
> or four days working consequatively) to get the initial major refactor
> done.  Once that's out of the way, we'll have a buggy, but working
> system where: 
> 
> 1. bmesh has replaced editmesh, for editmode and the ***EM functions
> in modifiers.
> 2. editmesh tools will still work, by way of using the accessor
> functions I made.
> 3. mesh dna will still be unchanged, I'm planning on doing that later.
> 4. derivedmesh will be unchanged.  IMHO, derivedmesh needs to be
> completely redone,
>     but no one has the time, so most likely we'll end up fixing the
> parts the most need
>     it, add face iterators so ngons work in e.g. subsurf, etc.  but
> I'm currently not planning
>     on doing this myself anytime soon. 
> 
>>
>> Which mesh DNA changes are needed? I thought f-gons flags would be
>> sufficient at that point?
> 
> Fgons rely on tesselations, which are a geometric property of the
> mesh.  Relying on geometry to accurately recreate topology causes
> issues, since deformed ngons may disappear if they can't be
> tessellated correctly.  We also can't use scanfill because of this,
> and the algorithm we are currently using (while relatively stable) is
> really slow. 
> 
> Currently I'm thinking of doing both; mfaces will store the
> tesselation, while an additional layer will store the original faces. 
> 
> Joe
> _______________________________________________
> Bf-committers mailing list
> Bf-committers at blender.org
> http://lists.blender.org/mailman/listinfo/bf-committers