Hello, thank you for the review and sorry indeed for anything we have
messed up.

I think the biggest issue is that we we were not aligned with the SRU
process here, and we thought that the specific cases of SRU'ing
backporting an entire version would be okay like that.

So, as the docs mention:
> Any deviance from the ideal SRU must be called out and justified

We've tried to do that, but it was probably not clear enough. This package is 
used by CUDA maintainers (us) to work on CUDA. It required some adjustements 
for future versions to come. 
- cuda-crt-13-1 is blocked in proposed because of this (and it's rather a 
critical one in regards of CUDA)
- all cuda 13-2 SRUs, currently in the unapproved queue will also benefit from 
this SRU
- all cuda 13-3 SRUs, yet to be submitted after 13-3 is accepted in stonking, 
will also benefit from this SRU

And Andreas said in the SRU template:
> Cherry-picking does not make sense, because all of these additions will
   be required for the future CUDA 13-2 SRU

That meant "We would like to backport the version from stonking
entirely, instead of disassembling the update in patches". We require
the same version than in stonking anyway, so backporting seems safer and
easier for everyone.

We don't have a SRU exception for this. We have a drafted one [0] but it
was said at some point that the SRU team would prefer to see some actual
SRUs before delivering an exception, so that there is more context. That
make sense.

That being said, it'd seem right to link the several bugs that this
upload fix. We should have thought of that. However writing 4 SRU
templates instead of one does not feel right. What do you think?

Taking an SRU from autopkgtests as example [1], I did reference LP bugs
in there but did not write more than one SRU template. I did not use
`v$VERSION` which I just learnt about, thank you for the tip. What would
be the implication of using that? Would it be the same as just reporting
everything in one changelog entry, but in a cleaner automated way?

If you still prefer that we open multiple SRU template we can do it, but
I'm confused in understanding why we would need that. I would rather
test all relevant bugs in this SRU template (which is not the case in
the SRU template as I read it now).

[0] 
https://github.com/ubuntu/ubuntu-project-docs/pull/517#discussion_r3008253236
[1] https://bugs.launchpad.net/ubuntu/+source/autopkgtest/+bug/2144629

-- 
You received this bug notification because you are a member of Ubuntu
Bugs, which is subscribed to Ubuntu.
https://bugs.launchpad.net/bugs/2164908

Title:
  SRU: Update to upstream 1.0.4

To manage notifications about this bug go to:
https://bugs.launchpad.net/ubuntu/+source/ubuntu-cuda-packaging-tools/+bug/2164908/+subscriptions


-- 
ubuntu-bugs mailing list
[email protected]
https://lists.ubuntu.com/mailman/listinfo/ubuntu-bugs

Reply via email to