GitLab

* Reorganise GHC and GHC.Packages modules to get rid of import cycles
* Move "GHC.Packages" to "Packages", merge "GHC" into "Settings.Default"

* Drop ghcCabal package
* Merge GhcCabal and CabalFlags builders
* Use more consistent naming
* Resolve import cycle
* Rename parseCabal to parseCabalFile
* Remove outdated references to ghc-cabal from comments

This fixes #654.

There are only two important changes.

1) The first one fixes missing dependencies:

```diff
-pkgDependencies = fmap (fmap PD.dependencies) . readPackageDataFile
+pkgDependencies = fmap (fmap (map pkgName . packageDependencies)) . readCabalFile
```

Here `PD.dependencies` returned versioned package names, e.g. `ghc-8.7`, which then failed to match with non-versioned package names such as `ghc` in `contextDependencies`. Switching from `PD.dependencies` to `packageDependencies` fixes this.

2) I clearly remember that we didn't have this bug before, so I added some tests for our package dependency infrastructure to prevent such regressions in future:

```haskell
testDependencies :: Action ()
testDependencies = do
    putBuild "==== pkgDependencies"
    depLists <- mapM (pkgDependencies . vanillaContext Stage1) ghcPackages
    test $ and [ deps == sort deps | Just deps <- depLists ]
    putBuild "==== Dependencies of the 'ghc-bin' binary"
    ghcDeps <- pkgDependencies (vanillaContext Stage1 ghc)
    test $ isJust ghcDeps
    test $ pkgName compiler `elem` fromJust ghcDeps
    stage0Deps <- contextDependencies (vanillaContext Stage0 ghc)
    stage1Deps <- contextDependencies (vanillaContext Stage1 ghc)
    stage2Deps <- contextDependencies (vanillaContext Stage2 ghc)
    test $ vanillaContext Stage0 compiler `notElem` stage1Deps
    test $ vanillaContext Stage1 compiler `elem`    stage1Deps
    test $ vanillaContext Stage2 compiler `notElem` stage1Deps
    test $ stage1Deps /= stage0Deps
    test $ stage1Deps == stage2Deps
```

Everything else are cosmetic changes, fixing minor issues in comments, and adding TODOs. To figure out the failure in #654 I had to read some code I didn't write and my hands were automatically fixing some style inconsistencies with the rest of the Hadrian code base. (I'd like to emphasise that I make no judgement about which style is better, it's merely an attempt to make the code base look more homogeneous, which I think is useful.)

* Use Cabal directly in place of ghc-cabal; make build root configurable.

This commit implements two significant changes (that were not easy to
separate):

- Don't use ghc-cabal anymore for getting information about Haskell packages.
  We now instead directly use Cabal-the-library.

- Make the build root configurable. This effectively gets rid of the inplace
  logic and allows us to place _all_ build artefacts in some directory of
  our choice, by passing '--build-root <some path>' to hadrian.

The code for this was mostly taken from #445.

* fix documentation rules

* remove some leftover unrelated, commented-out code

* more documentation fixes, address some feedback

* cleanup

* more cleanup

* boot and configure explicitly in travis CI scripts

* update cabal/ghc versions in .travis.yml (8.0.x not supported anymore)

* temporarily disable dynamic ways in Settings.Default

* update appveyor script

* travis: when booting with 8.2.2, build a complete stage2 compiler

* Fix CI?

Try to fix the CI by adding the `debug` rts way back in.

* Update Quickest.hs

Replicate the make build systems build flavours.

* Update .travis.yml

- Run selftest, and build in separate instances.
- try with python2
- and unify mac to stage2

* Update .travis.yml

upgrade python on mac

* [travis] os x: test the freshly built ghc

* Get rid of two unused GhcCabalMode constructors

* fix ghc-split rule, get rid of Install/Wrappers rules

* address more feedback

* ConfiguredCabal -> PackageData, more comments, more feedback addressed

* make the complete stage 2 build the default

* use a dummy package instead of base in Rules.hs

* update CI scripts

* attempt at fixing hadrian's -c option

* .travis.yml: use -c everywhere again

* travis: back to explicit './boot && ./configure'

* update README.md and doc/user-settings.md to reflect configurable build root

* some more feedback

* move a bunch of types into dedicated modules

* address review feedback

* do away with Hadrian.Builder.Mode for now

See #403

See #403

* Move buildRoot into the Action monad, so it can be configured from command line in future

* Move settings from Setting.Path to Base and Context

* Simplify build rule matching and prepare to factoring out common build rules into the library, #347

See #347

See #347

See #347

See #347

See #347

See #347

See #244.

See #207.

See #207.

See #188.

[skip ci]

Refactor getSingleton to avoid using list, otherwise an import of Data.Extra.List causes shadowing issues

The former is much more common than the latter.

The beginning of a long journey towards minimal re-exports