Move deprecated evm versions to nightly CI workflow
As recommended in the discussion here: https://github.com/ethereum/solidity/pull/14705#discussion_r1407879318, this PR moves the soltest to run exclusively on nightly jobs for Homestead and Byzantium EVM versions, which are now deprecated.
@cameel is more of an expert for the CI intricacies here than I am, so maybe we let him give final approval here - I'd merge it myself :-).
One more thing. Maybe you could also create a separate nightly-ossfuzz workflow and move t_ubu_ossfuzz to it? This way we could re-enable failure notifications for the other nightly jobs. Currently we're ignoring them all because ossfuzz always fails so won't really notice if they start failing too.
I'd do that in a separate PR though.
One more thing. Maybe you could also create a separate
nightly-ossfuzzworkflow and movet_ubu_ossfuzzto it? This way we could re-enable failure notifications for the other nightly jobs. Currently we're ignoring them all because ossfuzz always fails so won't really notice if they start failing too.
Right. But isn't it a matter to just add the step matrix_notify_failure_unless_pr to all nightly jobs except t_ubu_ossfuzz? Why do we need a new workflow? Some of the nightly jobs even already have such step.
Oh, so we still have gitter notifications for some of this?
I was thinking more about the failure notification e-mail I get from nightly every day. Now I have to always check it to see if there were any other failing jobs, which is mildly annoying. If ossfuzz was in a separate workflow, I could just add a filter to ignore it and only care about notifications about the other workflow (which I would now only be receiving when it really fails).
I mean, if we had consistent gitter notifications for all nightly jobs maybe I could ignore these mails altogether. But my impression was that not all jobs have them.
Oh, so we still have gitter notifications for some of this?
I was thinking more about the failure notification e-mail I get from nightly every day. Now I have to always check it to see if there were any other failing jobs, which is mildly annoying. If ossfuzz was in a separate workflow, I could just add a filter to ignore it and only care about notifications about the other workflow (which I would now only be receiving when it really fails).
I mean, if we had consistent gitter notifications for all nightly jobs maybe I could ignore these mails altogether. But my impression was that not all jobs have them.
I see, I don't know why I don't receive such notifications though, it might be some github config on my side, I will check it. Anyway, done here: https://github.com/ethereum/solidity/pull/14831
I see, I don't know why I don't receive such notifications though, it might be some github config on my side, I will check it.
I think it's in CircleCI settings. There's a page that lists the repositories you have access to and for each one you can select if you follow it. Then CircleCI sends you a notification about failed workflows. I get them for any PR where I pushed some commits and for the nightly runs.