No cabal file found for Archive

[waddlaw]

Steps to reproduce

1. create new project.

$ stack new sample
$ cd sample

2. edit stack.yaml (add extra-deps)

$ cat stack.yaml
resolver:
  url: https://raw.githubusercontent.com/commercialhaskell/stackage-snapshots/master/lts/12/9.yaml

packages:
- .
extra-deps:
- github: fpco/store
  commit: 247f2863bed0f5c1003ab835075b6dcafb9aa773

3. build project

$ stack build
No cabal file found for Archive from https://github.com/fpco/store/archive/247f2863bed0f5c1003ab835075b6dcafb9aa773.tar.gz

Some packages will succeed.(ex. cassava)

extra-deps:
- git: https://github.com/haskell-hvr/cassava.git
  commit: 151f74b976ce93c59e938ae130324bec53ac8c92
# - github: fpco/store
#   commit: 247f2863bed0f5c1003ab835075b6dcafb9aa773

Stack version

$ stack --version
Version 1.10.0, Git revision 388232c3f3ae350e384b2e41c0c11606fd06b69e (6449 commits) x86_64 hpack-0.30.0

[brandon-leapyear]

✨ This is an old work account. Please reference @brandonchinn178 for all future communication ✨

---

Also breaks if you include

subdirs:
- .

Error is in Pantry.Archive stripping prefix of a path instead of stripping subdirectories. e.g. specifying

subdirs:
- foo/bar

in a repository with the path foo/bar-asdf.cabal, it will resolve this as -asdf.cabal

[waddlaw]

@brandon-leapyear @mihaimaruseac
thanks! But, still unsolved problem...

I think that parseArchive is cause.

https://github.com/commercialhaskell/stack/blob/master/subs/pantry/src/Pantry/Archive.hs#L392

...
(cabalPath, cabalEntry@(TreeEntry cabalBlobKey _)) <- findCabalFile pli tree
...

With this implementation, it's necessary .cabal file in the archive. (don't auto generate from package.yaml)

We must to contains .cabal files in archive, even if exist package.yaml?

[mihaimaruseac]

Reopening in that case, the file should be regenerated

[quasicomputational]

This is in the same region as #2255 and #4267.

[snoyberg]

Please see #4377.

[snoyberg]

Fixed by #4439