Page MenuHome
Differential D5755 Diff 18321 source/blender/blenkernel/intern/particle.c

Changeset View

Standalone View

View Options

source/blender/blenkernel/intern/particle.c

Show First 20 Lines • Show All 2,910 Lines • ▼ Show 20 Lines if ((psys->edit == NULL || pset->flag & PE_DRAW_PART) == 0) {
return; return;
} }
} }
keyed = psys->flag & PSYS_KEYED; keyed = psys->flag & PSYS_KEYED;
baked = psys->pointcache->mem_cache.first && psys->part->type != PART_HAIR; baked = psys->pointcache->mem_cache.first && psys->part->type != PART_HAIR;
/* clear out old and create new empty path cache */ /* clear out old and create new empty path cache */
psys_free_path_cache(psys, psys->edit); /* another suggestion, much more general (and risky?) */
psys_free_path_cache(psys, psys_orig_edit_get(psys));
sergeyUnsubmitted
Not Done
Inline Actions

You are freeing cache for an original psys, but allocating it for evaluated one?
Can't be right.

sergey: You are freeing cache for an original psys, but allocating it for evaluated one? Can't be right.
lichtwerkAuthorUnsubmitted
Not Done
Inline Actions

During DEG evaluation:

  • patch frees the original of the PTCacheEdit->pathcache
  • code then allocates for the evaluated ParticleSystem->pathcache, true. [ not looking at this atm., might be another can of worms ;) ]

But regarding the PTCacheEdit->pathcache [ I am only looking at this atm ]:

  • apparently it is neccessary to free the original above, so this works during drawing later [drw_particle_update_ptcache > drw_particle_update_ptcache_edit > PE_update_object > psys_cache_edit_paths]
    • allocation for the new PTCacheEdit->pathcache then happens in psys_cache_edit_paths (if it was free above)

not sure what that tells us?

  • PTCacheEdit only lives on the original? OK...
  • anything that should behave differently? If it lived on the evaluated as well, would it have to be flushed back or something?

Anyways:

  • (a) still think second suggestion should be quite safe? [from what I can tell, and it fixes other things as well...T69000]
  • (b) but could also go with the first suggestion [tried and tested], would then do the same stuff for T69000 again...

What do you think to get this finally of the table: (a) or (b)?
Thanx a whole lot for looking at this btw., I know there is probably more urgent stuff to do.
(I promise not to make many more detours into hair anymore -- even though I dont see this being replaced any time soon?)

lichtwerk: During DEG evaluation: - patch frees the **original** of the `PTCacheEdit->pathcache` - code…
cache = psys->pathcache = psys_alloc_path_cache_buffers( cache = psys->pathcache = psys_alloc_path_cache_buffers(
&psys->pathcachebufs, totpart, segments + 1); &psys->pathcachebufs, totpart, segments + 1);
psys->lattice_deform_data = psys_create_lattice_deform_data(sim); psys->lattice_deform_data = psys_create_lattice_deform_data(sim);
ma = give_current_material(sim->ob, psys->part->omat); ma = give_current_material(sim->ob, psys->part->omat);
if (ma && (psys->part->draw_col == PART_DRAW_COL_MAT)) { if (ma && (psys->part->draw_col == PART_DRAW_COL_MAT)) {
copy_v3_v3(col, &ma->r); copy_v3_v3(col, &ma->r);
} }
▲ Show 20 Lines • Show All 2,000 Lines • Show Last 20 Lines