Haskell Foundation DevOps Weekly Update, 2023-02-15

Hello, it’s the weekly log!

I’m still putting most of my effort into removing spurious failures from GHC CI. A few smaller tasks filled out my available time. I wish I had more time in the day!

Finished:

  • Looked into another batch of spurious failures
  • Got set up with password sharing among maintainers of various Haskell services
  • Made improvements to the configuration for the GHC GitLab server, which was necessary to streamline deploying my spurious failure bot
  • Spent another small amount of time on Cabal pipelines on GitLab

Ongoing:

  • (Administrative stuff, taking care of filing US taxes as a resident of Finland)

Coming up:

  • Continuing with spurious failures
  • Describe this “spurious failure” process a little better. I will write a post about what the goal is, what it isn’t, how it fits into the bigger picture for the Haskell Foundation, and what the results have been so far.
6 Likes

Is there a link to see what changed about Cabal pipelines on GitLab?

1 Like

Oh, nothing has changed yet, I’m just exploring the GitLab CI code. I understand that Cabal releases may or may not be built on GitHub instead of GitLab soon. I’m actually not paying too much attention to that debate, and I’m not spending very much time on my exploration at all. :slight_smile:

3 Likes

@artem: I asked Bryan on #hackage if he could have a quick look at cabal CI and I OKed with HF (@david-christiansen) that diverting an epsilon of Bryan’s attention to that is acceptable and that syncing or not syncing release platforms of GHC and cabal (and other tools? HLS?) is a question worth exploring and asking the community/communities about.

That “quick look” was ostensibly to see in what shape the cabal gitlab CI is (though I knew very well it was in a terrible shape) and to estimate how costly it would be to get it and keep it fully in synch with GHC CI in the cheapest way possible, so that the community has any data to base its gitlab/github decision on. But, in reality, the release of cabal 3.10 is at hand and I secretly hoped Bryan is going to will cabal CI into working order just by the sheer threat that if it doesn’t Bryan may lay his hands on it. Or that there’s a script add-tool-to-GHC-CI.sh and Bryan runs it with argument cabal and that’s ti.

BTW, the discussion regarding cabal (but would it better to have a wider discussion?) may be started at [META] Regarding the release CI runners: GitHub or the Haskell GitLab? · Issue #8674 · haskell/cabal · GitHub once people have time to devote to it and if more than one option is still on the table by that time.

5 Likes

That is an invalid comparison:

  • Haskell GitLab is based on www.haskell.org - a solitary self-hosted instance;

  • Haskell GitHub relies on GitHub proper, and and the thousand-and-one servers it runs on.

It’s “apples and oranges”

1 Like

Maybe my first post here
i use lambda bot on irc among other and ways to use it are simple
i am still not clear about which link /documentation to follow, before i start using spurious bot ( as per the email today, from here)
Am i missing something?

Hey, I guess you are in the GHC contributors workshop, and the email I sent to that group is the one you are talking about.

Spuriobot isn’t something you use directly. It runs on the gitlab server.