Eelco Chaudron <[email protected]> writes:

> On 27 Jan 2026, at 17:40, Aaron Conole via dev wrote:
>
>> While adding new code, we can sometimes miss whether there is sufficient
>> test case coverage that accommodates the new functionality.  Additionally,
>> code that lacks sufficient test case coverage is ripe for new
>> contributors to find ways of delivering meaningful contributions to the
>> code base.
>>
>> Add a new CI job that executes the tests with code coverage and archives
>> the results.  To help facilitate this, update the coverage check to also
>> add '-fprofile-update=atomic' (due to some code paths being
>> multi-threaded) and make the github job explicitly not optimize code.
>> The resulting jobs can be analyzed for more details offline due to the
>> archives.
>>
>> Signed-off-by: Aaron Conole <[email protected]>
>> ---
>
> Hi Aaron,
>
> Thanks for working on this. I’m a big fan of adding coverage testing,
> but I want to make sure we get the maximum benefit from the extra CI
> time this will consume.
>
> My main concern is that without a pass/fail threshold, the data might
> be ignored. Do you have a vision for how contributors would use these
> logs effectively, especially given GitHub's limited artifact
> retention?

The idea would be to generate a report that we could use to understand
whether code that was modified actually got tested.

> It would be nice if we could eventually get to a "Diff Coverage"
> report (highlighting only the new code), though I know that is
> technically complex with our current diff/patching setup (similar to
> clang-analyze).

Yes, but that's the goal for the end.

> For now, do you think we should hold off on merging this until we can
> add some enforcement or clearer reporting? I’m worried about
> increasing the CI load without a guaranteed actionable outcome.

Okay by me.  I don't think it should increase the CI load so much, but I
guess it can make sense to shelve this work.

> //Eelco

_______________________________________________
dev mailing list
[email protected]
https://mail.openvswitch.org/mailman/listinfo/ovs-dev

Reply via email to