TSC Meeting Notes 2022-02-10

Owned by Cary Phillips
Feb 12, 2022
2 min read

Attendance:

Cary Phillips
Christina Tempelaar-Leitz
John Mertic
Joseph Goldstone
Kimball Thurston
Larry Gritz
Nick Porcino
Peter Hillman
Rod Bogart
Lucas Miller

Discussion:

  • For the Open Source Forum, we need:

  •  

    • High-level executive summary, “ROI”

    • Some performance stats from studios/applications starting to use the new core library 

  • Branch rename:

  •  

    • Cary: will do it this weekend, so it can be reported at the OSF.

  • Kimball: I’ll run clang format, will make a push, not a PR since it touches every line.

  • Lucas: I thought it was going to be about multiple imaths

  • Kimball: I started converting the c++ layer to use core. Immediately ran into problems. C layer uses c structs for vecs. Why doesn’t Imath use those as well?  Could add them. Only for when Imath is header only. Will submit a draft PR for discussion.

  • Larry: Been tying myself in knots about multiple Imaths. But here is an attempt to solve it, with a shim class: https://github.com/OpenImageIO/oiio/pull/3330. Struct v2f, then the Imath::V2f then derives from the data storage. It’s just for parameter passing

  • Alembic uses Imath types in its public API

  • We might be able to propagate these changes to other libraries

  • Kimball: we’ve talked about not having a version in the namespace

  • Larry: But what would happen if you decided you want your V3’s padded to 4 for simd alignment? 

  • Joseph:

  •  

    • The fight to have decent standard attributes continues.

    • SMPTE is trying to not be jammed up in paperwork.

    • ILM didn’t want OpenEXR to be tied up in SMPTE paperwork.

    • Possibility of leveraging SMPTE’s reputation.

    • They have a github repo as a way of providing visibility on proposals.

    • That’s where I would go to propose updates to ImfStandardAttributes.h

    • Kimball: gotta run, but this is what I think we were suggesting before (having a wiki that is controlled via PR), go Joseph, sounds great to me!

    • Nick: the github repo would be a public wiki scratch pad.

    • Joseph: the repo would force public debate, not just anyone can merge. 

    • “We have to wait for the results of the VES metadata survey to come back.”

    • Not going to be an instantaneous thing.

    • But will eventually come back to us.

    • Today was the first discussion, it’s ongoing.

  • Peter: one outstanding fuzz issue: out of memory. Will limit the number of samples.