GitLab

E.g. jaeger-flamegraph.cabal failed to roundtrip 'pretty . parse';

Making the list business work didn't feel right,
so I made a bit bigger refactor.

It serves an almost identical purpose as extra-library-flavours, except
that the latter is only concerned with static library flavours, while
extra-dynamic-library-flavours lets us, in the case of libHSrts, build
a dynamic, threaded flavour of it.

Create a new syntax for depending on any library of any package.
The syntax is

    build-depends: pkgname:{pkgname, sublibname} -any

where the second `pkgname` specifies a dependency on the main unnamed
library.

Closes #4206.

Previously, when running `cabal check` all the various ghc-*-options
flags were merged together, thus losing the information about the exact
place of the warning. This PR implements separate checking of
ghc-*-options, which allows us to give users more precise warnings.

Fixes #5342

This has been a problem since #5372 began expanding globs in `cabal
check`. Now the logic of running a glob is separated from the parsing,
giving the caller the opportunity to handle parsing failures flexibly.

These globs make `haddock`, `sdist` and `install` die, so they
definitely ought to be warned about!

Because of the dying behaviour, I made the checks dist-inexcusable;
starting from a clean slate I would probably have only made them
suspicious, but this isn't terrible.

This also significantly improves the error when trying to refer to
missing directories, hopefully making it clear that it's coming from
Cabal. #5318 and snowleopard/hadrian#634 are two bugs which manifested
as Cabal trying to glob in a non-existent directory and both took some
debugging because of the obscurity of the error.

Previously, we were checking the package with a hard-coded root
directory of ".". This was not a problem before, but with #5372 we
have started to expand globs while checking packages, which breaks if
the CWD is not the directory containing the `.cabal` file and causes
snowleopard/hadrian#634.

Luckily, this is an easy fix: the correct directory is easy to
determine. Writing a test and making sure it's tickling the failing
case took longer than writing the fix!

"." is hard-coded as the root directory passed to `checkPackageFiles`
in a few other places, but those are (a) non-trivial to test, and (b)
already in places that have other assumptions about their CWD, so I
have simply documented the CWD requirement for those.

This has the effect of allowing a glob `*.html` to match the file
`foo.en.html`. For compatibility, this is only allowed with
`cabal-version: 3.0` or later; for earlier spec versions, a warning
will be generated by `cabal check` if there are files affected by this
change in behaviour.

Fixes #5057. Fixes #784. Closes #5061.

These are inspired by a plan described in a comment in #2522, and only
implement a quite limited form of recursive matching: only a single **
wildcard is accepted, it must be the final directory, and, if a **
wildcard is present, the file name must include a wildcard.

Or-patterns are not implemented, for simplicity.

Closes #3178, #2030.

It's 'extra-source-files', not 'extra-src-files'.

(cherry picked from commit a6c69f25)

Follow-up to 2a375443

[skip ci]

Just enough to simplify 'Distribution.PackageDescription.Configuration'
for now.

Fixes #5086

The https://github.com/haskell/cabal/pull/5054 links to
https://github.com/commercialhaskell/stack/issues/3789 which says

- `Ensure you have OverloadedStrings and RebindableSyntax extensions
  enabled.`

So we warn only in that case. Only `OverloadeStrings` (or
`OverloadedLists`) or `RebindableSyntax` seems to be ok.

Also make `allBuildInfos` return all (not only buildable) build infos,
removing FIXME. `allBuildInfos` is used only in D.PD.Check.

 - Post flag-solving, replace buildDepends with solved dependencies.

 - In most cases, just collect the build-depends of *enabled* components.
   B4ut in a few, we have no enable spec on hand and must collect them
   all

Solves #5087
Related to #5003

Note: `buildable: ` field has `All` semantics. `scope: ` has
`AnyPrivate` semantics. Both might be surprising, but are "logical".

https://hackage.haskell.org/package/Cabal-2.0.1.1/docs/src/Distribution.PackageDescription.Parse.html#line-248

Resolves #2547

I introduce SimpleLicenseExpression to make "OrAnyLater LicenseRef"
unrepresentable. That also simplifies types.

license field is parsed as old `License` when cabal-version <2.2,
and as SPDX expression otherwise. `NONE` is recognised.

There are best-effort functions to convert between `License` and
`SPDX.License`.

There are also IPI changes: parser accepts both `License` and
`SPDX.License`, as both can occur in package database. Cabal will
`register` a SDPX expression as `license` for GHC >= 8.4, and legacy
`License` for other (we are smart when converting `PackageDescription` +
LBI and other data to `InstalledPackageInfo`)

Also add NFData InstalledPackageInfo

This is a follow-up to #4958 which opened up the opportunity to do
this make-illegal-states-unrepresentable refactoring as well.

This is to address #4551.

This implements the following defaulting rules:

 * For `cabal-version:2.0` and below, default to the `Custom`
   build-type unconditionally (legacy defaulting)

 * Otherwise, if a `custom-setup` stanza is defined, default to
   the `Custom` build-type; else default to `Simple` build-type.

This gets us better defaults for the two most popular use-cases, and
which can be statically inferred by only looking at the `.cabal` file.

This allows us to bring down the minimal (modern) trivial cabal
package definition down to a single file with 4 lines:

    cabal-version: 2.1
    name: mu
    version: 0
    library

NB: We don't need any `Setup.hs` file, as `cabal sdist` will magically
generate one on the fly.

This refactoring unifies the defaulting logic into a single place
paving the way for changing the defaulting logic.

[ci skip]

Add --enable-split-sections flag and pipe it through to the GHC backend. Note
that some of the implementation here could be made a bit more precise:
-split-sections and -split-objs are mutually exlusive yet the types don't
currently reflect this.

Fixes #4819.

Properly list 'extra-library-flavours' instead of 'virtual-modules' for `extraLibFlavours`.

Add `checkVersion` for asm-sources, cmm-source, extra-bundled-libaraires, extra-library-flavours, and virtual-modules.

As pointed out by @hvr, #4857, #4875 did not contain the necessary "check" logic. This PR tries to address this shortcoming.