Why is this redundant looking call required?

[Robadob]

When adding support for keyframe animated models, I removed the below line of code thinking it was a mistake. However, this lead to the keyframe model's vertex attribute buffers appearing empty (unless the Shaders object was reloaded). When I added this line back it started working again.

As far as I could tell at a glance, this line is doing nothing that would benefit from it being called twice (the second call to addGenericAttributeDetail() passes update=true, only a single call should be required to update both).

FLAMEGPU2-visualiser/src/flamegpu/visualiser/shader/Shaders.cpp

Line 517 in 31cefa0

buildVAO();

Doesn't seem hugely important to understand the odd behaviour, but could be useful to clearly document (or fix) to save the line being inadvertently removed in future.

[MILeach]

I haven't tried to reproduce this but it sounds like very weird behaviour. I can't see any object state being mutated within this function. The only thing that really makes sense to me here is that some of the preceding code is leaving OpenGL in an invalid state causing something to fail on the first run through. The invalid state is restored to a valid one by the buffer binds etc. taking place within the function so then it succeeds second time round.

[Robadob]

the preceding code is leaving OpenGL in an invalid state causing something to fail on the first run through. The invalid state is restored to a valid one by the buffer binds etc. taking place within the function so then it succeeds second time round.

Mm, will need to dig out that OpenGL calls debugger to try and catch that.

[MILeach]

Could also have a try with Renderdoc if you haven't done that already, very good for debugging graphics stuff

[Robadob]

Can't remember which tool I previous used. I think it was a dll replacement, which generated a text file with all of the calls to openGL in order.

…

On Wed, 17 Aug 2022 at 13:42, MILeach ***@***.***> wrote: Could also have a try with Renderdoc if you haven't done that already, very good for debugging graphics stuff — Reply to this email directly, view it on GitHub <#98 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AAFVGCULFNDSKWSPR537AEDVZTM4FANCNFSM56ZIYTGQ> . You are receiving this because you authored the thread.Message ID: ***@***.***>