[Bf-committers] Do we accept to have infinite recursive RNA layout patterns?

[Kévin Dietrich]

Hi Bastien, 

I remember reading some details/documentation about this situation when
I worked on supporting OpenVDB in the PointCache system, but I can't
find the source back. The only thing I found is a line from Lukas' wiki
page about his Alembic point cache ideas [1]: 

"Furthermore the RNA definition for these point cache lists is horrible:
it defines a collection property inside PointCache, which leads back to
the list of which it is a part - uagh! Presumably that was a lazy
solution to avoid defining the collection in each of the point cache use
cases ..." 

So the culprit seems to be our good old friend laziness. Maybe there is
a more detailed explanation out there. 

Cheers, 

Kévin 

Le 2018-03-18 16:04, Bastien Montagne a écrit :

> FYI, decided to go with a somewhat ugly, but safe and non-API-breaking solution for now, see https://developer.blender.org/rB0301df40e5b6c51575d7f9013a1a28b901063829
> 
> Bastien
> 
> On 16/03/2018 15:50, Bastien Montagne wrote: 
> 
>> Hi devs,
>> 
>> So, was investigating a weird issue today found while doing some static override tests, and ended up finding the someone, at some era lost in the mist of times, had the genius idea of creating a sort of infinite recursion in  PointCache RNA struct (defined in rna_object_force.c). That is, pointcache owner has a pointer to active pointcache, and said active pointcache has a collection of all pointcaches from owner's list (since in DNA, PointCache is actually a linked list item).
>> 
>> Needless to say that this totally breaks attempt to walk in RNA data, since you'll indefinitely dive into recursive versions of point_caches collection items (giving RNA paths like that: 'particle_systems["feathers_big"].point_cache.point_caches[0].point_caches[0].point_caches[0]....').
>> 
>> Before nuking this non-sense away and sanitizing the mess, I thought I'd ask if someone remembers any good reason for current RNA code/layout there... Since changing that will introduce RNA API breakage (though point_cache does not seems to be used by any addon in our repositories currently), would also probably only do it in 2.8 branch?
>> 
>> Cheers,
>> Bastien
>> 
>> _______________________________________________
>> Bf-committers mailing list
>> Bf-committers at blender.org
>> https://lists.blender.org/mailman/listinfo/bf-committers
> 
> _______________________________________________
> Bf-committers mailing list
> Bf-committers at blender.org
> https://lists.blender.org/mailman/listinfo/bf-committers

 

Links:
------
[1]
https://wiki.blender.org/index.php/User:Phonybone/PointCache/InitialThoughts