1 patch for repository http://darcs.net/screened:

patch 872a8c26404e56b0a31944c340e8c6a0254bd66a
Author: Ben Franksen <ben.franksen@online.de>
Date:   Sun May 16 14:19:52 CEST 2021
  * Cabal-Version: 2.2 -> 2.4
  
  This removes `cabal check` warnings caused by extra-source-files wildcards.

Attachments

• cabal_version_-2_2-__-2_4.dpatch
• patch-preview.txt
• unnamed

fine

This appears to be breaking CI:
https://github.com/bfrk/darcs-ci/runs/2651426149

I'm rather confused as according to the docs, the Cabal-Version is synced
with the actual Cabal version:
https://cabal.readthedocs.io/en/3.4/file-format-changelog.html#package-
description-format-specification-history

There's a reference to
https://github.com/haskell/cabal/issues/4899
which I haven't digested properly yet.

Note this potentially affects the 2.16.4 release as well.

> This appears to be breaking CI:
> https://github.com/bfrk/darcs-ci/runs/2651426149

Yikes!

> I'm rather confused as according to the docs, the Cabal-Version is synced
> with the actual Cabal version:
> https://cabal.readthedocs.io/en/3.4/file-format-changelog.html#package-
> description-format-specification-history

Very strange: the problem exists only with ghc-8.4.4. How is that possible?

> There's a reference to
> https://github.com/haskell/cabal/issues/4899
> which I haven't digested properly yet.
> 
> Note this potentially affects the 2.16.4 release as well.

It does. Should I roll back and make a new release?

> Very strange: the problem exists only with ghc-8.4.4. How is that possible?

No idea. Also I am trying to reproduce it and haven't managed to yet: so
far I am trying GHC 8.4.4/Cabal 3.2.0.0 on NixOS (nixpkgs commit dd6ce8c).

>> Note this potentially affects the 2.16.4 release as well.
> 
> It does. Should I roll back and make a new release?

Let's wait a bit and try to characterise the problem first.

> No idea. Also I am trying to reproduce it and haven't managed to yet: so
> far I am trying GHC 8.4.4/Cabal 3.2.0.0 on NixOS (nixpkgs commit dd6ce8c).

I take that back: I can reproduce it with that environment. Turns out it
only failed after building dependencies.

I'll dig further now.

Found it, see patch2181.

I guess we ought to fix the 2.16 release too as it is likely to hit
anyone running GHC 8.4.

> Found it, see patch2181.

Thanks a lot for digging into this (and finding the proper solution)!

> I guess we ought to fix the 2.16 release too as it is likely to hit
> anyone running GHC 8.4.

Agreed. I would like to include the fix for issue2677 (not yet sent), in
the new release, while we're at it.