21st September 2020
- Ashley Whetter - Chairperson (DNEG)
- David Aguilar (Disney Animation)
- Dora Morolli (Weta)
- Ryan Bottriell (Imageworks)
- Daniel Heckenberg (Animal Logic)
- Sam Assadian (Isotropix)
- Zhicheng Ye (Animal Logic)
- JT Nelson (Pasadena Python)
- Mark Wiebe (AWS)
- Sean Wallitsch (DWA)
- Robert Vinluan (SideFX)
- Welcome to isotropix
- Sam from Isotropix
- Review action points for last meeting
- Any comments for current Github pull requests
- List of backport modules
- No comments at this time.
- What is blocking people from doing the python 3 conversion?
- General Discussion
-
Getting dates from vendor when Python3 is coming, and when Python2 will be dropped.
-
JT: Do we wants names of affiliations on the PRs?
- Ashley: If we have information it would make sense to include.
-
What is blocking python3 / progress?
- Ashley: About halfway with the core team converting
- Ryan: Moving slowly. Metrics setup, and warnings being reported. No way to track python bindings, but not part of the report. Every few weeks someone donverts their project but there is struggle to make progress on the conversion front. Actively re-evaluating deployment toolchain (spfs) and how we launch and manage software. Don't want to tie the two projects together. Aware of work, but not much interest from studio/production at the moment.
- Dora: Weta is in a similar situation. A lot of legacy code not on python3, but if it is not updated then everyone gets blocked. Not having a DCC ready blocks because it doesn't push priorities. It's easier for projects that have no dependencies; easy to get blocked by internal dependency blockage.
- Ashley: Everything is being questioned as to whether it should be updated. VFX platform support is wide; looking to close off the older end before extending support on the upper end.
- Zhingcheng: Animal Logic is working towards porting to 2+3 compatibility. DCC is the main blocker. Slow progress. Using linters to reduce variants in the source, and using wheels to deploy. Using future and six. Nohthing too fancy; just standard tools.
- Daniel: including USD in the DCC-agnostic work. Building upon the python3 support in USD.
- Ryan: not much chance to test the python3 versions.
- Robert: SideFX will have a supported 18.5 release with python3. Tested for quite a while with alpha/beta users. R&D team testing all tools in py3. Houdini has done a lot of testing. A supported python3 build be available to the public soon.
- Sean: have tried to keep py3 compatibility in mind for a while, but testing has not been extensive. Rez package work underway. New stuff on python3, but not there yet. Working on a vfx2019 platform with python3 swapped out for python3.
- Ryan: Cannot match numpy versions in Houdini across python versions because the older version is not py3 compatible, so the 2 vs. 3 houdini foundation will differ in numpy versions.
- Mark: always working on things in a future py3-compatible approach.
- Sam: Clarisse finished py3 port from about 6 months ago. No push from studios, especially with WFH concerns from studios. Goal is to have the same build where python2 and python3 can be brought up in the same application. Symbol clash issues across platforms, so the user has to specify either py2 or py3 on startup. Should be released before the end of the year. vfx2020 is the target.
- David: similar situation, not much progress in the near term. Knowing when DCCs will drop py2 support will push production towards adoption.
- Sam: next version w/ python3 support will be the last version with py2. Next couple of years, but that's it.
- Robert: Houdini 18.5 will have py3 available, but py2 is default. Houdini 19 will have py2 available, but py3 will be default. Would like to drop py2 after that, TBD on studio needs.