Per https://github.com/haskell/cabal/wiki/Updating-the-license-list
to version 3.23 (2024-02-08) of SPDX License List.

Substitute “internal library” occourrences with “sublibrary”.

`cabal-version: 2.4` allows e.g. `data-files: foo/**/*.txt` but `**`
directories are only handled correctly since Cabal 3.0 (issue #6125).
This information should be part of the cabal file format history
documentation.

Closes #8069

+ Update file-format-changelog.rst

* allow glob matches of the form dir/**/FileNoExtension

* allow glob matches of the form dir/**/FileNoExtension

* tests, docs, changelong

Co-authored-by: Gershom Bazerman <gershom@arista.com>
Co-authored-by: mergify[bot] <37929162+mergify[bot]@users.noreply.github.com>

* wip to add test-code-generators field to test stanzas

* fixups

* change hashes

* regen golden parser test output

* docs and changelog

* test

* Update pr-7688

* tweak test

Co-authored-by: Gershom Bazerman <gershom@arista.com>

Niklas Hambüchen authored 5 years ago and Alex Biehl committed 3 years ago

* WIP: Still need to update ghc to use the new `libraryDirsStatic` and
       `extraLibrariesStatic` fields in `InstalledPackageInfo` for linking.
* WIP: Adding -L paths to .a libs in .conf files will likely increase nix
       closure size for dynamically linked Haskell packages because if pkg-config
       finds an `-a` file, the path to it will be remembered.
       (That only has an impact if the .a file is in a split-output from the .so file.)
       Perhaps we make the feature that `pkg-config --static` is called
       opt-outable.
       On the other hand Haskell libs always carry their .a files, so pulling
       a few system `.a` files more shouldn't make much of a difference.
       So this opt-out should probably be an extra feature for later, if at all.

Fixes #4761

Signed-off-by: Edward Z. Yang <ezyang@fb.com>

ghstack-source-id: 957ba1db25747ffb08ad8b9a4af1f9310d8e738d
Pull Request resolved: https://github.com/haskell/cabal/pull/7306

Simon Marlow authored 5 years ago and Oleg Grenrus committed 4 years ago

Motivation: I (simonmar) want to use hsc2hs with C++, which requires

```
hsc2hs-options: --cc=g++ --lflag=-lstdc++
```

We pass `hsc2hs-options` before or after `--cc` and `-lflag`,
depending on the version of `hsc2hs`.

I (phadej) added a tests:
- hsc2hs-options are passed to `hsc2hs` (but not to GHC e.g.).
- `--cc` has expected effect

`g++` tests doesn't seem to work on Windows, due mingw ncurses linkage
issue, https://github.com/msys2/MINGW-packages/issues/3531
We also need non-ancient hsc2hs (>= 0.68).

- Add availableSince to language/extension fields
   - Also autogen-modules once we are on that.
- Remove default-language check for cabal-version: 3.4

We cannot make the field `uniqueFieldAlaDef`, as that would require
specifying `default-language` in each subconditional.
I leave it for https://github.com/haskell/cabal/issues/6925

Note: the default language is whatever is the compiler's default.
The GHC-8.10 default is its variant of Haskell2010
with NondecreasingIndentation and without DatatypeContexts

Herbert Valerio Riedel authored 5 years ago and Oleg Grenrus committed 5 years ago

While those buildinfo fields were added to the parser some time
ago via 57d7f28e and
4a287659 that work was never completed
by implementing the necessary build/sdist logic in Cabal.

This commit remedies this oversight by implementing and wiring up the
missing build logic.

*WARNING* this commit is still very WIP; ASM works mostly; C--
          support is still incomplete

autogen-includes aren't searched nor packages by `sdist`.

This is relatively small code patch, but there are
- change in file-format
- short documentation of the field
- `cabal check`
- test-suite noise due new field in `BuildInfo`

Starting with `cabal-version: 3.0` free text fields
preserve indentation and blank lines.
In other words, we don't need single dots to indicate blank line.

IPI is not versioned. So we default to latest variant always.
This is little harm, as old GHC use old Cabal to print
IPIs. And free text fields are only informative.

Fixes #5938

- Don't accept leading zeros in Version or VersionRange
- Fixes https://github.com/haskell/cabal/issues/163
  (pkg-config uses a more general version scheme)
- Also resolves https://github.com/haskell/cabal/issues/5138
- `PkgconfigVersion` is compared with `rpmvercmp`
- `PkgconfigVersionRange` is subset of `VersionRange`
    - with `cabal-version` before 3.0 it's parsed like `VersionRange`,
      where version digits are arbitrary integral (leading spaces
      allowed)
    - starting from cabal spec 3.0 `== x.y.*` and `^>=`
      (and set `{ ..  }`) are disallowed.
      Yet, the version literals syntax is relaxed to accept
      alphanumerical + `-.` strings. E.g. openssl's `1.1.0h` is
      accepted.
- Lax `PkgconfigVersion` parser is also used to parse
  `pkg-config --modversion` output.

- extensions
- (singular) hs-source-dir
- build-tools

Example, with `cabal-version: 2.5`

```
[laptop] tree-diff % cabal new-build --dry
Errors encountered when parsing cabal file ./tree-diff.cabal:

tree-diff.cabal:59:3: error:
The field "extensions" is removed in the Cabal specification version 2.5. Please use 'default-extensions' or 'other-extensions' fields.

   58 | library
   59 |   extensions: CPP
      |   ^

tree-diff.cabal:60:3: error:
The field "extensions" is removed in the Cabal specification version 2.5. Please use 'default-extensions' or 'other-extensions' fields.

   59 |   extensions: CPP
   60 |   extensions: DeriveGeneric
      |   ^
```

I needed to add new CabalSpecVersion to properly issue deprecation
messages of `extensions`.