Deadline support for {representation} anatomy key

[BigRoy]

@timsergeeff

Need some info do we need outputdirectory0 env in deadline task? Looks like ayon ignores it at publishing stage because it publishes fine with litteraly brocken path whitch i’ve created using my “fix” i think maybe we can remove this code and this env and it will publish just fine

That is Deadline metadata which allows you to browse the directory directly from the Deadline browser - it’s not crucial. But it’s far from the only entry point that may rely on publishDir not being PER representation as I’ve said before.

1. Integrator: ayon-core/client/ayon_core/plugins/publish/integrate.py at 3eb960a8568e5366ad685572c9107c653b8fb9ed · ynput/ayon-core · GitHub
2. Publish dir and resource dir instance data, used for e.g. resources for maya look or yeti publishes: ayon-core/client/ayon_core/plugins/publish/collect_resources_path.py at 95980b10f32d59d645c81ef0dc38d2e9903e4936 · ynput/ayon-core · GitHub
   • e.g. yeti: ayon-maya/client/ayon_maya/plugins/publish/extract_yeti_cache.py at 06aa2445b4df511225b6795b283dd974040c0b5f · ynput/ayon-maya · GitHub
3. Hero version integration: ayon-core/client/ayon_core/plugins/publish/integrate_hero_version.py at f61675b10e85c309c33a794fd5911c45af788425 · ynput/ayon-core · GitHub
4. Defining relative USD files when publishing before the integration (and hence potentially before knowing all representations): ayon-core/client/ayon_core/plugins/publish/extract_usd_layer_contributions.py at 065033a9244cee9e00f56d84f66f2e178a39adc1 · ynput/ayon-core · GitHub
5. Maya Obj publish relative files: ayon-maya/client/ayon_maya/plugins/publish/extract_obj.py at 972dc58db9734fa24bc84649744780d2b7d9cc31 · BigRoy/ayon-maya · GitHub
6. Houdini USD look publish relative texture file copy on publish: ayon-houdini/client/ayon_houdini/plugins/publish/collect_usd_look_assets.py at ed2a8018350c0c19ad6b63af9180abf73a919f43 · ynput/ayon-houdini · GitHub

And so forth.

I really do think it is not trivial to support this without some massive overhaul. And even then, I think it will make things way more complex. Some file publishes need to remap paths inside an actual exported file without knowing yet where the exact representation lives where it’s reading from. Whether there’s a way to design it wherein that’s trivial to solve I’m not sure. But it’d require a lot of changes, and thorough testing. I wonder if the development time spent on it to make sure all features work is worth it for what it ends up solving on the publish paths themselves.