AY-6199_Resolve: Ayon menu not loading

ynbot commented 3 weeks ago

Please describe the issue you have acountered?

Ha - damned Resolve. I've heard of Resolve or Fusion 'crashing' their python interpreter with OpenTimelineIO but so far only on Linux... with 0.17.0 also on Windows that's news for me.

[16:38] @jakubjezek001 @Murphy (ynput)Definitely something we should track as issue so that the logs are clearer for the script Force a print statement before the imports to ensure we know at least the script run started (for future debugging) Somehow safeguard the import of OpenTimelineIO Preferably report if a "known invalid" OpenTimelineIO version is installed (if we can do that without importing it, because importing it.. crashes it) | of course to get there we may want to first reproduce the crash on our end to with 0.17.0 to confirm) Improve the README and installation instruction for Resolve to state to install OpenTimelineIO 0.15.0 to avoid further debugging sessions like this. A part of this is also mentioned here:

How to replicate the issue?

No response

Additional context:

link to discussion on Discord (might be a private channel)

This issue was automatically created from Clickup ticket AY-6199

jakubjezek001 commented 6 days ago

Ideal solution for this would be to have own prelaunch hook for OpenTimelineIO dependency installing - actually the same way as it is done for example in Flame here https://github.com/ynput/ayon-flame/blob/develop/client/ayon_flame/hooks/pre_opentimelineio_install.py

BigRoy commented 6 days ago

Ideal solution for this would be to have own prelaunch hook for OpenTimelineIO dependency installing - actually the same way as it is done for example in Flame here https://github.com/ynput/ayon-flame/blob/develop/client/ayon_flame/hooks/pre_opentimelineio_install.py

Just want to mention that ideally we find a better generalized way for distributing across the board - because e.g. Blender is slow to start because it needs to check Blender's python on each launch whether it was installed or not. Same goes for an issue I had for C4D (where its c4dpy is extremely slow to launch for whatever reason, so the query took 10s+) (issue here). Preferably we can do indeed roll out this via any way a studio prefers and maybe do a light-weight check to see if e.g. something is deployed or not - or find other better solutions or whatever. However that may grow it out of scope of this particular issue and throw it into @dee-ynput 's lap for an "Epic" down the line.

But for now - if the check is quick, a prelaunch hook could suffice. Yes.

ynput / ayon-resolve