Also rename CONTRIBUTORS to AUTHORS, which seems to be more standard.

[ci skip]

Report process output decoding errors in context

Not all the exception handling, just for the case fixed by the previous
patch.

Turns out of course that this is sensitive to the locale encoding, so we
just skip the test when the localeEncoding is not UTF-8.

Adjustments for what deps we pick for legacy Setup scripts in new-build

Add a lower bound of >= 1.18. Also improve the related comments.

Using >= 1.18 since we currently have problems working with 1.16 as that
version lacks support for named build targets. We ought to fix working
with older versions though.

Add transformers and old-time, and for ghc-compatible compilers also add
ghc-prim and template-haskell.

Have to double check if the old-time dep needs to be conditional, e.g.
on ghc version.

It turns out that in rawSystemStdInOut any IO errors, including text
decoding, occurring while collecting the output (stderr or stdout) do
not get reported immediately, but only later if/when the output is
consumed. This is because the exceptions happened in the threads forked
to force the output, but if the main thread looks at the output later
then the exception is reported again.

This leads to very confusing results. In particular we had a case where
the configure process did not fail until writing out the LocalBuildInfo
because that was the first point that forced the output. So the distance
from when the exception really occurred and the fact that the exception
message does not include the name of the program run means that this is
then a pain to track down.

This patch makes sure that any exceptions arising from forcing the
program output occur immediately and with an error message that includes
the name of the program in question.

Signed-off-by: Edward Z. Yang <ezyang@cs.stanford.edu>

Unfortunately, it was too difficult to factor out the common code
between ProjectBuilding and Install.

Signed-off-by: Edward Z. Yang <ezyang@cs.stanford.edu>

Signed-off-by: Edward Z. Yang <ezyang@cs.stanford.edu>

Signed-off-by: Edward Z. Yang <ezyang@cs.stanford.edu>

Move solver types to Distribution.Solver.* namespace

Randy Polen authored 8 years ago and Mikhail Glushenkov committed 8 years ago

contents. Prior to the response file code, all of these
details of the command line for haddock would have been
logged (with level == Verbose), so this corrects an oversight
and brings the information in the logging back to what it
used to be for cabal's haddock invocations.

* Cabal/Distribution/Simple/Haddock.hs
  * Restore the logging of the entire command line used when
    invoking haddock to what it used to be prior to adding
    the response-file creation.

(cherry picked from commit 28bea0d7)

Signed-off-by: Edward Z. Yang <ezyang@cs.stanford.edu>

Simplify `GoalChoice`

It makes the build fail due to timeout.

/cc @ezyang

[ci skip]

This commit looks somewhat bigger than it is because I moved some comments
around so we can have longer comments for the constructors. The most important
change is the documentation of the `QGoalReason` invariant on 'GoalChoice'.

This commit doesn't change any actual code.

OpenGoal contains a FlaggedDep which contains a lot more information than we
actually need in the tree; we just need to know, which choice are we
introducing: package, flag, or stanza? We don't need the full FlaggedDep tree.

Pkg-config constraints are actually a part of 1.24.

Add `conflictSetOrigin` debugging field

Make `Validate` a proper monad

Add missing exports

This field is only added if the `debug-conflict-sets` flag is specified to
`cabal config`; it adds a tree to `CallStack`s to a `ConflictSet`. This is very
useful when trying to understand how a certain conflict set was constructed.

This makes it a bit easier to modify it (for example, during debugging).