Exclude broken 'Validate windows-latest ghc-8.4.4' job that cancels CI

[Mikolaj]

See #8115 (comment) and later comments.

[jneira]

lgtm but what about some brief comment about?

[andreabedini]

Sorry I have been out of the loop, is this a temporary thing or permanent? if permanent I suggest leaving a comment that explains why that configuration is missing. Otherwise someone silly like me will show up one day and re-add it :D

[Mikolaj]

@andreabedini: hi! Permanent, see the comment at the link (copied from #hackage).

Is the exisitng comment # lot of segfaults caused by ghc bugs, and this PR that git blame finds, lacking?

Edit: @jneira ^^^

[jneira]

hmm my limited english is failing me sorry, do you mean the comment for ghc 8.8.4 is applicable to the new version exception? are we getting ghc segfaults for 8.4.4?

[Mikolaj]

Yes, that would be my guess given this happened for 8.8.4 and given we are getting "Access violation in generated code when reading 0xffffffffffffffff". Is it plausible?

[Mikolaj]

If by "ghc segfaults" you mean "segfaults caused by ghc bugs", not necessarily when running ghc, but possibly when running other binaries produced by ghc, as in our case.

[Mikolaj]

It merged already, but I will improve the comment in a new PR, if we decide it's worth it.

[jneira]

hmm an it also throws segfaults could be sufficient

[Mikolaj]

Sure. We currently have

        exclude:
          # lot of segfaults caused by ghc bugs
          - os: "windows-latest"
            ghc: "8.8.4"
          - os: "windows-latest"
            ghc: "8.4.4"
        include:

Where would you put "it also throws segfaults"?

[jneira]

Where would you put "it also throws segfaults"?

nah, between them, but it looks good as is, dont worry

[Mikolaj]

@jneira: thanks

@andreabedini: how about you?

[andreabedini]

I’m ok with this change but disabling all tests on windows on GHC 8.8.4 and 8.4.4 means we cannot claim to support that configuration. I think we are discussing this in some other thread but perhaps this is a reminder we need to have sour support story straight.

my 2c, good work team!

[Mikolaj]

That's a good point. We currently claim in CONTRIBUTING.md:

Our GHC support window is five years for the Cabal library and three years for cabal-install: that is, the Cabal library must be buildable out-of-the-box with the dependencies that shipped with GHC for at least five years. The Travis CI checks this, so most developers submit a PR to see if their code works on all these versions of GHC. cabal-install must also be buildable on all supported GHCs, although it does not have to be buildable out-of-the-box.

We could change "The Travis CI checks this" to "Our Github Actions CI setup checks this (except for a couple of GHC/OS combinations that crash too readily)". Any other place we need to touch?

[andreabedini]

@Mikolaj I just set to rephrase that paragraph and spent 30 minutes finding the GHC release dates 🤦 I guess the best we have is the blog.

[andreabedini]

I left a comment in #8108