GitLab

I discovered that in

   let x = MkT y in ....(MKT y |> co)....

we weren't CSE'ing the (MkT y).  The fix is easy.

Trac #14035 showed that -XStrict was generating some TERRIBLE
desugarings, espcially for bindings with INLINE pragmas. Reason: with
-XStrict, all AbsBinds (even for non-recursive functions) went via the
general-case deguaring for AbsBinds, namely "generate a tuple and
select from it", even though in this case there was only one variable
in the tuple.  And that in turn interacts terribly badly with INLINE
pragmas.

This patch cleans things up:

* I killed off AbsBindsSig completely, in favour of a boolean flag
  abs_sig in AbsBinds.  See Note [The abs_sig field of AbsBinds]

  This allowed me to delete lots of code; and instance-method
  declarations can enjoy the benefits too.  (They could have
  before, but no one had changed them to use AbsBindsSig.)

* I refactored all the AbsBinds handling in DsBinds into a new
  function DsBinds.dsAbsBinds.  This allowed me to handle the
  strict case uniformly

* Clarify the comments around the mc_strictness field of FunRhs
* Use record field names consistently for FunRhs

as requested in #14045.

[skip ci] comments only

Test Plan: ./validate on harbormaster

Reviewers: austin, hvr, bgamari, erikd, Phyx

Reviewed By: Phyx

Subscribers: Phyx, rwbarton, thomie

GHC Trac Issues: #12494

Differential Revision: https://phabricator.haskell.org/D3726

Summary:
Correct a couple more spots in the TH pretty-printer by applying the
appropriate parenthesization for infix names. Fixes #13887 (again).

Test Plan: make test TEST=T13887

Reviewers: austin, bgamari

Subscribers: rwbarton, thomie

GHC Trac Issues: #13887

Differential Revision: https://phabricator.haskell.org/D3802

Previously, we checked the number of patterns in a data instances
for all data families whose kind did not end in a kind variable.
But, of course, undersaturating instances can happen even without
the kind ending in a kind variable. So I've omitted the arity check.
Data families aren't as particular about their arity as type families
are (because data families can be undersaturated). Still, this change
degrades error messages when instances don't have the right arity;
now, instead of reporting a simple mismatch in the number of patterns,
GHC reports kind errors. The new errors are fully accurate, but perhaps
not as easy to work with. Still, with the new flexibility of allowing
data family instances with varying numbers of patterns, I don't see
a better way.

This commit also improves source fidelity in some error messages,
requiring more changes than really are necessary. But without these
changes, error messages around mismatched associated instance heads
were poor.

test cases: indexed-types/should_compile/T14045,
            indexed-types/should_fail/T14045a

This was proposed by David Feuer
(https://mail.haskell.org/pipermail/libraries/2016-August/027293.html)
and solves #14029.

The implementation is a copy of the '<&>' operator in Control.Lens.Lens.

Add tests for following Data.Functor operators: '<$>', '<&>', '<$' and '$>'.

'<&>' was added for solving #14029. '<$>', '<$' and '$>' were probably
untested.

Reviewers: austin, hvr, bgamari, RyanGlScott

Reviewed By: RyanGlScott

Subscribers: RyanGlScott, rwbarton, thomie

Differential Revision: https://phabricator.haskell.org/D3800

Test Plan: Validate, ensure the `f7` program of `IPLocation` doesn't
crash.

Reviewers: gridaphobe, austin, hvr

Reviewed By: gridaphobe

Subscribers: rwbarton, thomie

GHC Trac Issues: #14028

Differential Revision: https://phabricator.haskell.org/D3795

These ignore commandline arguments for ignore and commandline as well as
GHCRTS arguments for ignoreAll. Passing RTS flags given on the command
line along to the program by simply skipping processing of these flags
by the RTS.

This fixes #12870.

Test Plan: ./validate

Reviewers: austin, hvr, bgamari, erikd, simonmar

Reviewed By: simonmar

Subscribers: Phyx, rwbarton, thomie

GHC Trac Issues: #12870

Differential Revision: https://phabricator.haskell.org/D3740

Fixes #13710.
Signed-off-by: Edward Z. Yang <ezyang@cs.stanford.edu>

Test Plan: validate

Reviewers: bgamari, austin, simonpj

Reviewed By: simonpj

Subscribers: simonpj, rwbarton, thomie

GHC Trac Issues: #13710

Differential Revision: https://phabricator.haskell.org/D3743

This was an old workaround for #5252. Fixes #13173.

Reviewers: austin, bgamari

Reviewed By: bgamari

Subscribers: rwbarton, thomie

Differential Revision: https://phabricator.haskell.org/D3763

This is only supported by Python 3.5 and later, which is too new for us
to rely on.

Reviewers: austin

Subscribers: rwbarton, thomie, RyanGlScott

GHC Trac Issues: #14050

Differential Revision: https://phabricator.haskell.org/D3803

ld.gold is particularly picky that we declare all of our link
dependencies on Nix. See #14022.

Test Plan: Validate on Nix

Reviewers: austin

Subscribers: hvr, rwbarton, thomie

GHC Trac Issues: #14022

Differential Revision: https://phabricator.haskell.org/D3787

This broke in the fix for #13541.

If you have ld.gold and ld.lld, then ld.gold will be selected by the
detection logic. This patch prioritizes lld by changing the order. The
rationale for checking lld first is that it's (right now) not part of,
say, a default Linux distro installation and if it's available, it's
very likely that it was installed explicitly and should be seen as a
sign of preference. On FreeBSD LLVM is the (default) base toolchain and
the changed order makes sense there as well, since ld.gold can be
available in /usr/local via ports/pkg. I don't have access to macOS and
can't say anything about their LLVM toolchain.

At some point we could add a check for LD=ld.lld or LD=ld.gold as an
optional override to explicitly select a linker.

Since I cannot really remove gcc on Linux, this was the only way to
configure GHC to use ld.lld.

Reviewers: austin, hvr, bgamari

Reviewed By: bgamari

Subscribers: rwbarton, thomie, erikd

Tags: PHID-PROJ-5azim3sqhsf7wzvlvaag

Differential Revision: https://phabricator.haskell.org/D3790

This is another change needed for #13825 (also based on D38 by Simon
Marlow).

With the change, we count the stack depth in bytes (instead of words).
We also introduce some `newtype`s to help with the change.

Note that this only changes how `ByteCodeGen` works and shouldn't
affect the generated bytecode.
Signed-off-by: Michal Terepeta <michal.terepeta@gmail.com>

Test Plan: ./validate

Reviewers: bgamari, simonmar, austin, hvr

Reviewed By: bgamari, simonmar

Subscribers: rwbarton, thomie

GHC Trac Issues: #13825

Differential Revision: https://phabricator.haskell.org/D3746

This allows to customize the location where binary distributions are
placed with `make binary-dist`.

E.g. using:
```
BINARY_DIST_DIR=/path/to/bindists make binary-dist
```
will place binary dists outside of the source tree into the given
folder.

This change falls back to ".", which is the old behaviour.

Test Plan: build binary-dist

Reviewers: bgamari, austin

Reviewed By: bgamari

Subscribers: rwbarton, thomie

Differential Revision: https://phabricator.haskell.org/D3798

Reviewers: austin

Subscribers: rwbarton, thomie

GHC Trac Issues: #14028

Differential Revision: https://phabricator.haskell.org/D3788

Test Plan: Validate, try ingesting into Jenkins.

Reviewers: austin

Subscribers: rwbarton, thomie

GHC Trac Issues: #13716

Differential Revision: https://phabricator.haskell.org/D3796

Tangentially related to my prior work on trac ticket #12758.
Signed-off-by: Jared Weakly <jweakly@pdx.edu>

Reviewers: austin, bgamari

Subscribers: rwbarton, thomie

Differential Revision: https://phabricator.haskell.org/D3792

Summary:
Some recent commits happened to fix other issues:

* c2417b87 fixed #13601 and #13780
* 8e15e3d3 fixed the original program in #13877

Let's add regression tests for each of these to ensure they stay fixed.

Test Plan: make test TEST="T13601 T13780a T13780c T13877"

Reviewers: goldfire, bgamari, austin

Reviewed By: bgamari

Subscribers: rwbarton, thomie

GHC Trac Issues: #13601, #13780, #13877

Differential Revision: https://phabricator.haskell.org/D3794

Summary:
Types and kinds are now the same in GHC... well, except in the code
that involves Template Haskell, where types and kinds are given separate
treatment. This aims to unify that treatment in the `DsMeta` module.

The gist of this patch is replacing all uses of `repLKind` with `repLTy`.
This is isn't quite as simple as one might imagine, since `repLTy` returns a
`Core (Q Type)` (a monadic expression), whereas `repLKind` returns a
`Core Kind` (a pure expression). This causes many awkward impedance mismatches.

One option would be to change every combinator in `Language.Haskell.TH.Lib` to
take `KindQ` as an argument instead of `Kind`. But this would be a breaking
change of colossal proportions.

Instead, this patch takes a somewhat different approach. This migrates the
existing `Language.Haskell.TH.Lib` module to
`Language.Haskell.TH.Lib.Internal`, and changes all `Kind`-related combinators
in `Language.Haskell.TH.Lib.Internal` to live in `Q`. The new
`Language.Haskell.TH.Lib` module then re-exports most of
`Language.Haskell.TH.Lib.Internal` with the exception of the `Kind`-related
combinators, for which it redefines them to be their current definitions (which
don't live in `Q`). This allows us to retain backwards compatibility with
previous `template-haskell` releases, but more importantly, it allows GHC to
make as many changes to the `Internal` code as it wants for its purposes
without fear of disrupting the public API.

This solves half of #11785 (the other half being `TcSplice`).

Test Plan: ./validate

Reviewers: goldfire, austin, bgamari

Reviewed By: goldfire

Subscribers: rwbarton, thomie

GHC Trac Issues: #11785

Differential Revision: https://phabricator.haskell.org/D3751

Summary:
Previously, if `reifyInstances` failed to discover a `Name` during
renaming, it would blindy charge into typechecking, at which point GHC would
become very confused at the absence of that `Name` and throw an internal error.
A simple workaround is to fail eagerly after renaming errors.

Test Plan: make test TEST=T13837

Reviewers: goldfire, austin, bgamari, simonpj

Reviewed By: simonpj

Subscribers: simonpj, rwbarton, thomie

GHC Trac Issues: #13837

Differential Revision: https://phabricator.haskell.org/D3793

This was provoked by an ASSERT failure when debugging #14038,
but it's a godo idea anyway.

Trac #14000 showed up two errors

* In TcRnTypes.dropInsolubles we dropped all implications, which
  might contain the very insolubles we wanted to keep.  This was
  an outright error, and is why the out-of-scope error was actually
  lost altogether in Trac #14000

* In TcSimplify.simplifyInfer, if there are definite (insoluble)
  errors, it's better to suppress the following ambiguity test,
  because the type may be bogus anyway.  See TcSimplify
  Note [Quantification with errors].  This fix seems a bit clunky,
  but it'll do for now.

This ASSERT failure (in substTy) was reported in Trac #14024.

This patch gets the in-scope set right.

(Does not fix tests T13822 or T13594.)

In Check.hs (pattern match ovelap checking) we to figure out the
instantiation of a pattern synonym from the type of the pattern. We
were doing this utterly wrongly.  Trac #13768 demonstrated this
bogosity.

The fix is easy; and is described in PatSyn.hs
  Note [Pattern synonym result type]

Summary:
Get utf8 encoded arguments before we call hs_init and use them
instead of ignoring hs_init arguments. This reduces differing
behaviour of the RTS between windows and linux and simplifies
the code involved.

A few testcases were changed to expect the same result on windows
as on linux after the changes.

This fixes #13940.

Test Plan: ./validate

Reviewers: austin, hvr, bgamari, erikd, simonmar, Phyx

Subscribers: Phyx, rwbarton, thomie

GHC Trac Issues: #13940

Differential Revision: https://phabricator.haskell.org/D3739

With the changes caused by the fix to #12369, it is now clearer
how to rewrite tcInferApps and friends. This should change no
behavior, but it does clean up a nasty corner of the type checker.
This commit also removes some uses of substTyUnchecked.

Previously, a data family's kind had to end in `Type`,
and data instances had to list all the type patterns for the
family. However, both of these restrictions were unnecessary:

- A data family's kind can usefully end in a kind variable `k`.
  See examples on #12369.

- A data instance need not list all patterns, much like how a
  GADT-style data declaration need not list all type parameters,
  when a kind signature is in place. This is useful, for example,
  here:

    data family Sing (a :: k)
    data instance Sing :: Bool -> Type where ...

This patch also improved a few error messages, as some error
plumbing had to be moved around.

See new Note [Arity of data families] in FamInstEnv for more
info.

test case: indexed-types/should_compile/T12369

Previously, looking up a TyCon that said "no" to mightBeUnsaturated
would then instantiate all of its invisible binders. But this is
wrong for vanilla type synonyms, whose RHS kind might legitimately
start with invisible binders. So a little more care is taken now,
only to instantiate those invisible binders that need to be (so that
the TyCon isn't unsaturated).

This "fixes" #11995.

This is a straightforward fix -- there were just some omitted
checks.

test case: typecheck/should_fail/T11963

A type equality error can arise from a mismatch between
*invisible* arguments just as easily as from visible arguments.
But we should really prefer printing out errors from visible
arguments over invisible ones. Suppose we have a mismatch between
`Proxy Int` and `Proxy Maybe`. Would you rather get an error
between `Int` and `Maybe`? Or between `*` and `* -> *`? I thought
so, too.

There is a fair amount of plumbing with this one, but I think
it's worth it.

This commit introduces a performance regression in test
perf/compiler/T5631. The cause of the regression is not the
new visibility stuff, directly: it's due to a change from
zipWithM to zipWith3M in TcUnify. To my surprise, zipWithM
is nicely optimized (it fuses away), but zipWith3M is not.
There are other examples of functions that could be made faster,
so I've posted a separate ticket, #14037, to track these
improvements. For now, I've accepted the small (6.6%) regression.