regularly failing CI processes
levitte at openssl.org
Fri Mar 24 07:45:05 UTC 2023
We do run the CIs (we have two CI systems running: Github Actions and
buildbot) on pull requests just as well as on regular commits to the
master and release branches.
The pull requests are usually not perfect from the get go, so the CIs
may fail on them until everything is fixed, and this might be, as you
say, a slew.
Among the CI tasks done on PRs is to check that the documentation
that's written is formatted according to a set of standards, and cover
what it's supposed to cover. Among others, we have checks that ensure
that all public symbols are documented, unless they're written up in a
set of files in util/. If the documentation isn't up to our standards,
that CI task will fail.
You can, BTW, run that very task yourself at any time (it might need
some specific program to be installed, I frankly don't recall for the
$ make doc-nits
On Thu, 23 Mar 2023 20:01:11 +0100,
Michael Richardson wrote:
> First, a thank you for the time from Thomas and Randall towards getting a
> minor documentation fix into the tree. I got a slew of messages today from
> the CI system about failures. Since this was a documentation fix, it should
> affect no code, so it's a bit concerning.
> My impression is that the CI tests have become too big, and too unstable for
> many months, and I just can't count on them being clean with a clean tree
> before I start work.
> I've been through this with other projects, and I wonder if all these CI
> tests are truly providing value, or if some of them have just become noise
> that distracts the developers.
> [2 signature.asc <application/pgp-signature (7bit)>]
> No public key for 808B70FBDDD0DD65 created at 2023-03-23T20:01:11+0100 using RSA
Richard Levitte levitte at openssl.org
OpenSSL Project http://www.openssl.org/~levitte/
More information about the openssl-users