Attendees - Rod Burns (Codeplay) Chairperson - Robert Cohn (Intel) - Vinesh Sukumar (Qualcomm) - Andrew Wafaa (Arm) - Dave Murray (Imagination Technologies) - Penporn Koanakatool (Google) - Masahiro Doteguchi (Fujitsu) - Ramesh Radhakrishnan (VMware) - Jory Burson (Linux Foundation) - Alison Richards (Intel)
Agenda - Operational Procedures Final Review - Working Group Priorities - Goals and metrics for 2024
Operational procedures are published as draft with final chance for review. A release branch has been created on GitHub and this defines the current operational procedures for the foundation. The members will continue to review and update these when required.
The groups previously discussed what the priorities should be for the two working groups - Specification and Open Source. A general summary of these was presented and discussed.
The Steering Committee will publish a set of more detailed priorities on the GitHub repository for the Working Groups.
The groups will meet on a regular basis, the intention initially is for the Open Source Working Group to meet every two weeks, and the Specification Working Group will decide the frequency when they meet.
Specification Priorities
- Project documentation for contribution and collaboration
- Roadmap for releases
- Best practices for community feedback on new APIs and designs
The Specification Working Group is planning the migration of the specification rendered content to GitHub pages from Intel internal infrastructure.
The group will also prioritise the alignment of how changes are shared for review. Some projects have a RFC process and others a variation on this.
A roadmap is already under discussion and will be published in due course.
Open Source Priorities
- Build & Infrastructure
- Open Source Best Practice
- Multi-Vendor Targets
- Open Source Kernels
- Architecture
- Compatibility Tests
Discussion points:
The structure of this group will need to be agreed by the members, it may be necessary to split this into multiple sub-groups as well as work packages.
Members can help to bring credits to the foundation for cloud environments that can be used for build and testing of any of the projects. (e.g. AWS, Google Cloud, MS Azure...)
The TensorFlow project provides the ability for commmunity members to hook their own infrastructure into the project to build and test on specific targets. This is something we should encourage for UXL projects. For example Intel already has a set of infrastructure for building and testing. Other members and contributors should be able to add their own.