The glitch and assert seem to be fixed. The Cycles error unexpected smoke volume resolution seems to be unrelated: it just cannot find the "temperature" grid, and the message is a bit confusing.…
I see that the cache is placed in a /tmp/ folder - is this just the OS removing the temp folder arbitrarily? I can't really reproduce it reliably.
I suppose the depsgraph just isn't specific enough to tell exactly under what circumstances a transform change of the emitter requires recalculation. In addition i think the particle/hair state is…
Can't tell what is happening here, but particles are not maintained, so will mark this a known issue.
Making this a known issue, particles are not maintained any more.
Seems to be working fine in latest main (4.3), confirmed in 2.93. Will close this.
The depsgraph currently just adds a relation between the driver node and the Parameters component. To trigger cache reset there would need to be a dependency from the parameters of any…
Probably a duplicate of #81423. Will mark this a known issue since mantaflow integration isn't actively maintained.
I can't reproduce this issue in 2.93, 4.1 or main (4.3). The callstack suggests it might generate invalid spring->particle_index in some race condition.
Since particles are not actively…
It would check if the attribute layer given as a parameter is actually owned by the drawing, and cancel if not (rather than try and find a layer with the same name).
