[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: ROCm 6.1: should we start uploading components to unstable?



Hi Christian,

Thank you for reviewing and uploading these updates.

On 2024-09-24 07:14, Christian Kastner wrote:
we have rocr-runtime, roct-thunk-interface, and amdsmi in experimental.
Should we start moving these to experimental?

amdsmi should be the least controversial, I think.

A cursory examination of our CI results did not show any issues with
rocr-runtime or roct-thunk-interface. If we update, which one should we
choose first? (I'd like to do a staggered update, just in case.)

We have good coverage of ROCm functionality in the continuous integration system. If the test results indicate that these packages are working, I have no concerns about uploading to unstable. In my opinion, mixing and matching package versions is of no concern as long as we have the test coverage to prove it works.

When uploading, I would start at the bottom of the dependency tree and move up. If nothing else, it slightly reduces the number of autopkgtests that will be scheduled for rdepends. That would be roct-thunk-interface first, rocr-runtime second and amdsmi third.

Sincerely,
Cory Bloor


Reply to: