Attendance:

Others:

Discussion:

Cary and Kimball discussed an upcoming release, a bug report, and the possibility of a patch release for the 3.3 version. They also discussed rewriting code logic for data unpacking, updating the build script and fuzzing system, and changing the name of the new compression codec to HTJ2K upon merging into the main branch. Lastly, they discussed progress on the Imath project, concerns about exr2aces functionality, and the process of vending the OpenJPH library.

Next steps

• Cary to start preparing release notes for OpenEXR 3.3.3 patch release.

• Kimball to implement fixes for issue #1991 and submit them over the weekend.

• Cary to leave a comment on PR #1975 explaining the plan to change the compression name to "HTJ2K" upon merging into main.

• Cary to prepare a release for Imath to include the fix for the undefined behavior bug.

• Cary to set up the Python wheel workflow for Imath to enable "pip install Imath".

• Peter to create an issue to revisit the exr2aces tool in light of future color management changes.

• Kimball to go through and close old or resolved issues in the project tracker, consulting with the team as needed.

• Kimball to work on vendoring the OpenJPH library into OpenEXR as the default behavior.

Summary

Upcoming 3.3 Release and Patch Discussion

In the meeting, Cary discussed the upcoming release and mentioned that Kimball needed to address an outstanding bug report from Phil Barrett. Kimball confirmed he hadn't looked into it yet and planned to do so. They discussed the possibility of a patch release for the 3.3 version, which Larry supported to avoid previous mistakes. It was decided that the next minor release should align with the VFX reference platform cut-off time.

Rewriting Code Logic for Data Unpacking

Kimball discussed rewriting the code logic for choosing optimal routines to unpack data, aiming to fix an issue. Cary agreed to delay the cut until this task is completed, planned for this weekend. Cary also proposed merging the HTJ 2K branch into the main branch before the next release, allowing time for testing and improvement.

PR Failing and Build Script Updates

Cary and Kimball discussed an issue with the PR failing in the fuzz test check and the need to update the build script and the fuzzing system. They also talked about adding an external app and incorporating dependencies into the system to avoid external downloads or links. Peter joined the discussion, and they discussed the potential security failure and the need for synchronizing the merge. They also discussed the possibility of reaching out to the fuzzing team to temporarily turn off the fuzzing while they work on the merge.

Change HTJ2K in Beta Branch

The group decided to change the name to HTJ2K upon merging into the main branch. They agreed to close PR 1975 and put a note on the Beta branch to capture this change. Peter suggested closing the ticket and asking the submitter to create a new one with the new name. Cary agreed to leave a comment on the PR after the meeting to communicate this plan.

Imath Project Progress

Cary reports progress on the Imath project, including moving confidence tests to the pybind11 binding and validating functionality for vector classes. He plans to prepare the Python wheel workflow, enabling "pip install Imath."

Documentation Discussion

The team discusses an open issue regarding documentation for exr to aces, with Peter explaining that the tool supports two extra compressions not in the standard, possibly due to historical reasons. They consider whether to document this discrepancy or fix the tool to match the standard.

exr2aces Deprecation and Pipeline Usage

Peter expressed concerns about exr2aces's functionality and suggested potentially deprecating it. Cary proposed either removing the tool or issuing a warning about its usage. Larry was open to adding the tool's functionality to OpenImageIO. Kimball suggested accepting the latest update, while Cary proposed going further and warning that the tool should not be used in pipelines.

ACES Tool Compatibility and Cleanup

The group discusses updating the exr2aces tool and its compatibility with new color management standards. Peter suggests accepting the current version and creating an issue to revisit the tool later when color management plans are clearer. Kimball proposes cleaning up old or resolved issues in the project tracker, and Cary encourages him to do so, suggesting to close any issues that are no longer relevant. The meeting ends with a brief mention of submitting something to OSS Fuzz to resolve an HT J2K fuzzing failure.

Vendoring in OpenJPH Library Process Discussion

In the meeting, Kimball, Peter, and Cary discussed the process of vendoring in the OpenJPH library. They agreed that the default behavior should be to vendor in the library, unless anyone has a strong opinion otherwise. They also discussed the possibility of having three different options for downloading, specifying a local path, or finding it on the system. Kimball mentioned that he made a script that auto updates the library and checks the upstream for newer versions. They also talked about the need to watch the tickets for potential changes. They concluded that they need to decide on the default behavior before the summer or fall release.