GitLab

Solver: Deduplicate flags and stanzas in DependencyReason.

The issue is fixed now.

Fix NondecreasingIndentation filtering

Improve `build-type` defaulting 

(see 9fb03d73 and #4958 for details)

This tweaks the existing `CustomPlain` test to test the legacy
defaulting logic for the unconditionally `Custom` default.

Morever, a new `SimpleDefault` test has been added which tests
that `cabal-version:2.1` does indeed infer `build-type: Simple`
when there is no `custom-setup` stanza defined.

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 commit is unlikely to have an effect on real packages, but it reduces the
chance of performance problems caused by duplicate dependencies in the solver
quickcheck tests.  It also adds a regression test to memory-usage-tests.

It is probably better to simplify version ranges in most cases in the solver, in
order to simplify error messages and save memory.  Currently, (.||.) is used once
for lifting dependencies out of conditionals, and (.&&.) isn't used.

This commit simplifies the solver's internal representation of package
conditionals by removing branches that are unreachable.  For example, the solver
now simplifies

if flag(A)
  if flag(A)
    build-depends: package-a
  else
    build-depends: package-b

to

if flag(A)
  build-depends: package-a

because A must be true for the second conditional to be relevant to the
finalized PackageDescription.

This change probably won't improve performance by simplifying real packages.
However, it fixes a bug in the solver's handling of a certain edge case.  Before
this change, it was possible for the solver to create DependencyReasons with
conflicting flag values.  The conflicting values could cause a problem when
one was FlagBoth, the value used when the same dependency appears on both sides
of a conditional.  Here is an example:

if flag(A)
  if flag(A)
    build-depends: unknown-package
  else
    build-depends: unknown-package

Previously, the solver inserted both A=FlagTrue and A=FlagBoth into the
DependencyReason for the dependency on unknown-package. When the previous
commit changed the association list of flag values into a map, FlagBoth replaced
FlagTrue and prevented the solver from trying -A.

This commit simplifies the conditionals to:

if flag(A)
  build-depends: unknown-package

It also adds some unit tests for dependencies occurring on both sides of
conditionals. "testBackjumpingWithCommonDependency" tests the example above.

This commit changes DependencyReason's list fields to maps and sets. Duplicate
flags were possible when a flag appeared multiple times in nested conditionals
or a flag controlled a "Buildable: False" field. The duplicate flag could show
up in log messages:

Before:
[__5] trying: json-rpc-client:+demo
[__6] trying: process-1.6.1.0/installed-1.6... (dependency of json-rpc-client +demo +demo)

After:
[__5] trying: json-rpc-client:+demo
[__6] trying: process-1.6.1.0/installed-1.6... (dependency of json-rpc-client +demo)

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

[skip ci]

RFC: Common stanzas

- common stanzas can be include other common stanzas
- `import: name1, name2` to import multiple stanzas
- Parse common stanzas in the same pass with other sections.
- Common stanzas have to be defined before use.
- Also negative tests
- Terse documentation, let's improve it as questions are asked

- Edit gen-extra-source-files to include golden files
- Amend elif warning to mention cabal-version: 2.2
- In regression golden tests, include also warnings

Note: ATM the common stanzas are completely handled inside parser,
GenericPackageDescription doesn't know about them anymore.
That can be changed, but the the flattening of
GenericPackageDescription to PackageDescription may fail.
I don't want to do that refactor now.

[ci skip]

Should unbork the test suite.

Not every supported extension for a compiler has a corresponding flag.
for example NondecreasingIndentation is enabled by default on GHC 7.0.4,
hence it is considered a supported extension but not an accepted flag.

To resolve this, wrap Flags in Maybe, and follow through the resulting
refactoring.

Fixes #4443

fixed comment typo: 'is' to 'it' [ci skip]

Fix cases where "it's" was used as a possessive

[ci skip]

This represents an incremental improvement over the current state.
However, there's more work to be done in terms of restructuring/reordering
to present the information in an easier to digest as well as apply the new
facilities effectively. This future work will be tracked in #4944.

[skip ci]

Foreign lib names not parsed by readp

travis-ci: bootstrap on 8.2.2 as well

Skip the Haddock step on macOS.

I don't think it ever uncovered any bugs that the Linux bot
missed. This should speed up the build a little.

[ci skip]

[ci skip]

[ci skip]

[ci skip]