GitLab

When using the :add command in haxlsh/ghci, a module/file that can't
be found is still added to the list of targets, resulting in an error
message for the bad module/file for every subsequent usage of the
command. The add command should verify that the module/file can be
found before adding it to the list of targets.

Also add a ":show targets" command to show the currently added list of
commands, and an ":unadd" command to remove a target.

Test Plan:
Add a new GHCi testcase that checks that :add doesn't remember either
files or modules that could not be found, and that both the new :show
and :unadd commands work as expected.

Reviewers: simonmar, bgamari

Reviewed By: simonmar

Subscribers: rwbarton, thomie, carter

GHC Trac Issues: #14676

Differential Revision: https://phabricator.haskell.org/D4321

Currently runtime keeps hold to 4*used_memory. This includes, in
particular, nursery, which can be quite large on multiprocessor
machines: 16 CPUs x 64Mb each is 1GB. Multiplying it by 4 means whatever
actual memory usage is, runtime will never release memory under 4GB, and
this is quite excessive for processes which only need a lot of memory
shortly (think building data structures from large files).

This diff makes multiplier to apply only to GC-managed memory, leaving
all "static" allocations alone.

Test Plan: make test TEST="T14702"

Reviewers: bgamari, erikd, simonmar

Reviewed By: simonmar

Subscribers: rwbarton, thomie, carter

GHC Trac Issues: #14702

Differential Revision: https://phabricator.haskell.org/D4338

Looks right to me.

Arguably the warning should just be disabled for this test to eliminate
unnecessary wiggle in the future.

These two tests have been failing on CircleCI.

Trac #14732 was a perpelexing bug in which -fdefer-typed-holes
caused a mysterious type error in a RULE.  This turned out to
be because we are more aggressive about creating implications
when deferring (see TcUnify.implicationNeeded), and the rule
mechanism hadn't caught up.

This fixes it.

We already prioritise equalities when solving, but
Trac #14723 showed that we were not doing so consistently
enough, and as a result the type checker could go into a loop.
Yikes.

See Note [Prioritise equalities] in TcSMonad.

Fixng this bug changed the solve order enough to demonstrate
a problem with fundeps: Trac #14745.

runTcSWithEvBinds does some unification, so the zonkWC
must be after, not before!  Yikes.  An outright bug.

This fixes Trac #14715.

- T12962: just fix function locations
- scc003: reorder cost centres, cost centre `f.(...)` renamed to `f.x'`

Reviewers: bgamari

Subscribers: rwbarton, thomie, carter

GHC Trac Issues: #14704

Differential Revision: https://phabricator.haskell.org/D4351

Currently, error messages that germane to GADT constructors
put the source span at only the first character in the constructor,
leading to insufficient caret diagnostics. This can be easily fixed
by using a source span that spans the entire constructor, instead of
just the first character.

Test Plan: make test TEST=T14719

Reviewers: alanz, bgamari, simonpj

Reviewed By: alanz, simonpj

Subscribers: simonpj, goldfire, rwbarton, thomie, carter

GHC Trac Issues: #14719

Differential Revision: https://phabricator.haskell.org/D4344

Previously `splitEnv` worked like this:

    > splitEnv "foo:::bar::baz:"
    ["foo","","","bar","","baz",""]

with this patch:

    > splitEnv working_dir "foo:::bar:baz:"
    ["foo",working_dir,working_dir"bar","baz",working_dir]

This fixes #14695, where having a trailing `:` in the env variable
caused ghci to pass empty `-B` parameter to `gcc`, which in turned
caused the next parameter (`--print-file-name`) to be considered as the
argument to `-B`. As a result ghci did not work.

The `working_dir` argument is to have a similar behavior with POSIX:
according to chapter 8.3 zero-length prefix means current working
directory.

Reviewers: hvr, bgamari, AndreasK, simonmar

Reviewed By: bgamari, AndreasK, simonmar

Subscribers: AndreasK, rwbarton, thomie, carter

GHC Trac Issues: #14695

Differential Revision: https://phabricator.haskell.org/D4330

This is an initial attempt at tackling the issue of how to order the
suggestions provided by the valid substitutions checker, by sorting
them by creating a graph of how they subsume each other. We'd like to
order them in such a manner that the most "relevant" suggestions are
displayed first, so that the suggestion that the user might be looking
for is displayed before more far-fetched suggestions (and thus also
displayed when they'd otherwise be cut-off by the
`-fmax-valid-substitutions` limit). The previous ordering was based on
the order in which the elements appear in the list of imports, which I
believe is less correlated with relevance than this ordering.

A drawback of this approach is that, since we now want to sort the
elements, we can no longer "bail out early" when we've hit the
`-fmax-valid-substitutions` limit.

Reviewers: bgamari, dfeuer

Reviewed By: dfeuer

Subscribers: dfeuer, rwbarton, thomie, carter

Differential Revision: https://phabricator.haskell.org/D4326

Subscribers: rwbarton, thomie, carter

Differential Revision: https://phabricator.haskell.org/D4334

Ideally, I'd like to do

    type EvTerm = CoreExpr

and the type checker builds the evidence terms as it goes. This failed,
becuase the evidence for `Typeable` refers to local identifiers that are
added *after* the typechecker solves constraints. Therefore, `EvTerm`
stays a data type with two constructors: `EvExpr` for `CoreExpr`
evidence, and `EvTypeable` for the others.

Delted `Note [Memoising typeOf]`, its reference (and presumably
relevance) was removed in 8fa4bf9a.

Differential Revision: https://phabricator.haskell.org/D4341

This gives a 10% allocation improvement on MultiLayerModules.
The idea is to reuse existing tuples, instead of constantly
constructing new ones.

Test Plan: ./validate

Reviewers: simonpj, bgamari

Reviewed By: simonpj, bgamari

Subscribers: rwbarton, thomie, simonmar, carter

Differential Revision: https://phabricator.haskell.org/D4332

Not doing so resulted in different uniques being printed on different
environments, as shown in #14703.

- Cabal-2.2 uses SPDX license identifiers, so I had to update
  `cabal-version: 2.1` packages `license: BSD3` to `license: BSD-3-Clause`
- `ghc-cabal` used old ReadP parsec, now it uses `parsec` too
- InstalledPackageInfo pretty-printing have changed a little,
  fields with default values aren't printed. This can be changed in
  `Cabal` still, but I haven't found problems with omitting them.

Note: `BSD-3-Clause` is parsed as "name = BSD, version = 3" by old
parser (because 3-Clause looks like version 3 with tag Clause).
If you see *"BSD-3" is not a valid license*, then something is using
old parser still.

Fixes #9885.

Before this change, a list was used as a substitute for a heap.
This led to quadratic behavior on a simple program (see new
test case).

This change replaces it with IntSet in effect reverting
5a1a2633. @simonmar said it's fine to revert as long as nofib
results are good.

Test Plan:
new test case:

20% improvement
3x improvement when N=10000

nofib:

I run it twice for before and after because the compile time
results are noisy.

- Compile Allocations:

```
          before    before re-run    after     after re-run
-1 s.d.   -----     -0.0%            -0.1%     -0.1%
+1 s.d.   -----     +0.0%            +0.1%     +0.1%
Average   -----     +0.0%            -0.0%     -0.0%
```
- Compile Time:

```
          before    before re-run    after     after re-run
-1 s.d.   -----     -0.1%            -2.3%     -2.6%
+1 s.d.   -----     +5.2%            +3.7%     +4.4%
Average   -----     +2.5%            +0.7%     +0.8%

```
I checked each case and couldn't find consistent slow-down/speed-up on
compile time. Full results here: P173

Reviewers: simonpj, simonmar, bgamari

Reviewed By: bgamari

Subscribers: rwbarton, thomie, carter, simonmar

GHC Trac Issues: #14667

Differential Revision: https://phabricator.haskell.org/D4329

Implement the proposal of underscores in numeric literals.
Underscores in numeric literals are simply ignored.

The specification of the feature is available here:
https://github.com/ghc-proposals/ghc-proposals/blob/master/proposals/000
9-numeric-underscores.rst

For a discussion of the various choices:
https://github.com/ghc-proposals/ghc-proposals/pull/76

Implementation detail:

* Added dynamic flag
  * `NumericUnderscores` extension flag is added for this feature.

* Alex "Regular expression macros" in Lexer.x
  * Add `@numspc` (numeric spacer) macro to represent multiple
    underscores.
  * Modify `@decimal`, `@decimal`, `@binary`, `@octal`, `@hexadecimal`,
    `@exponent`, and `@bin_exponent` macros to include `@numspc`.

* Alex "Rules" in Lexer.x
  * To be simpler, we have only the definitions with underscores.
    And then we have a separate function (`tok_integral` and `tok_frac`)
    that validates the literals.

* Validation functions in Lexer.x
  * `tok_integral` and `tok_frac` functions validate
    whether contain underscores or not.
    If `NumericUnderscores` extensions are not enabled,
    check that there are no underscores.
  * `tok_frac` function is created by merging `strtoken` and
    `init_strtoken`.
  * `init_strtoken` is deleted. Because it is no longer used.

* Remove underscores from target literal string
  * `parseUnsignedInteger`, `readRational__`, and `readHexRational} use
    the customized `span'` function to remove underscores.

* Added Testcase
  * testcase for NumericUnderscores enabled.
      NumericUnderscores0.hs and NumericUnderscores1.hs
  * testcase for NumericUnderscores disabled.
      NoNumericUnderscores0.hs and NoNumericUnderscores1.hs
  * testcase to invalid pattern for NumericUnderscores enabled.
      NumericUnderscoresFail0.hs and NumericUnderscoresFail1.hs

Test Plan: `validate` including the above testcase

Reviewers: goldfire, bgamari

Reviewed By: bgamari

Subscribers: carter, rwbarton, thomie

GHC Trac Issues: #14473

Differential Revision: https://phabricator.haskell.org/D4235

A silly mistake in `gen_Show_binds` was causing derived
`Show` instances for empty data types to case on the precedence
argument instead of the actual value being showed.

Test Plan: make test TEST=drv-empty-data

Reviewers: bgamari

Reviewed By: bgamari

Subscribers: rwbarton, thomie, carter

GHC Trac Issues: #14692

Differential Revision: https://phabricator.haskell.org/D4328

Subscribers: rwbarton, thomie, carter

GHC Trac Issues: #14670

Differential Revision: https://phabricator.haskell.org/D4314

This adds support for the bit deposit and extraction operations provided
by the BMI and BMI2 instruction set extensions on modern amd64 machines.

Implement x86 code generator for pdep and pext.  Properly initialise
bmiVersion field.

pdep and pext test cases

Fix pattern match for pdep and pext instructions

Fix build of pdep and pext code for 32-bit architectures

Test Plan: Validate

Reviewers: austin, simonmar, bgamari, angerman

Reviewed By: bgamari

Subscribers: trommler, carter, angerman, thomie, rwbarton, newhoggy

GHC Trac Issues: #14206

Differential Revision: https://phabricator.haskell.org/D4236

Subscribers: rwbarton, thomie

GHC Trac Issues: #14335

Differential Revision: https://phabricator.haskell.org/D4202

It turns out that `Convert` was recklessly leaving off
parentheses in two places:

* Negative numeric literals
* Patterns in lambda position

This patch fixes it by adding three new functions, `isCompoundHsLit`,
`isCompoundHsOverLit`, and `isCompoundPat`, and using them in the
right places in `Convert`. While I was in town, I also sprinkled
`isCompoundPat` among some `Pat`-constructing functions in `HsUtils`
to help avoid the likelihood of this problem happening in other
places. One of these places is in `TcGenDeriv`, and sprinkling
`isCompountPat` there fixes #14682

Test Plan: make test TEST="T14681 T14682"

Reviewers: alanz, goldfire, bgamari

Reviewed By: bgamari

Subscribers: rwbarton, thomie, carter

GHC Trac Issues: #14681, #14682

Differential Revision: https://phabricator.haskell.org/D4323

This reverts commit a1a689dd.

This implements SPJ's suggestion on the ticket (#14273). We find the
relevant constraints (ones that whose free unification variables are all
mentioned in the type of the hole), and then clone the free unification
variables of the hole and the relevant constraints. We then add a
subsumption constraints and run the simplifier, and then check whether
all the constraints were solved.

Reviewers: bgamari

Reviewed By: bgamari

Subscribers: RyanGlScott, rwbarton, thomie, carter

GHC Trac Issues: #14273

Differential Revision: https://phabricator.haskell.org/D4315

Trac #14646 happened because we forgot to parenthesize `forall` types to
the left of an arrow. This simple patch fixes that.

Test Plan: make test TEST=T14646

Reviewers: alanz, goldfire, bgamari

Reviewed By: alanz

Subscribers: rwbarton, thomie, carter

GHC Trac Issues: #14646

Differential Revision: https://phabricator.haskell.org/D4298

`LIBRARY_PATH` is used to find libraries and other link artifacts while
`LD_LIBRARY_PATH` is used to find shared libraries by the loader.

Due to an implementation detail on Windows, using `LIBRARY_PATH` will
automatically add the path of any library found to the loader's path.

So in that case `LD_LIBRARY_PATH` won't be needed.

Test Plan:
./validate along with T14611 which has been made Windows only
due to linux using the system linker/loader by default. So I feel a
testcase there is unwarranted as the support is indirect via glibc.

Reviewers: hvr, bgamari, erikd, simonmar, RyanGlScott

Reviewed By: RyanGlScott

Subscribers: RyanGlScott, rwbarton, thomie, carter

GHC Trac Issues: #14611

Differential Revision: https://phabricator.haskell.org/D4275

Distinguishing between "refutable" and "irrefutable" patterns
(as described by the Haskell Report) in incomplete pattern errors
was more confusing than helpful. Remove references to irrefutable
patterns.

Reviewers: hvr, bgamari, simonpj

Reviewed By: simonpj

Subscribers: simonpj, rwbarton, thomie, carter

GHC Trac Issues: #14569

Differential Revision: https://phabricator.haskell.org/D4261

Reviewers: bgamari

Reviewed By: bgamari

Subscribers: rwbarton, thomie, carter

Differential Revision: https://phabricator.haskell.org/D4256

This adds support for adding Haddocks on individual non-record fields
of regular (and GADT) constructors. The following now parses just fine
with `-haddock` enabled:

data Foo
  = Baz             -- ^ doc on the `Baz` constructor
      Int           -- ^ doc on the `Int` field of `Baz`
      String        -- ^ doc on the `String` field of `Baz`

  | Int             -- ^ doc on the `Int` field of the `:*` constructor
      :*            -- ^ doc on the `:*` constructor
    String          -- ^ doc on the `String` field of the `:*`
constructor

  | Boa             -- ^ doc on the `Boa` record constructor
      { y :: () }

The change is backwards compatible: if there is only one doc and it
occurs
on the last field, it is lifted to apply to the whole constructor (as
before).

Reviewers: bgamari, alanz

Subscribers: rwbarton, thomie, mpickering, carter

Differential Revision: https://phabricator.haskell.org/D4292

Before this change, for each constructor that we want
to allocate a tag for we would traverse a list of all
the constructors in a datatype to determine which tag
a constructor should get.

This is obviously quadratic and for datatypes with 10k
constructors it actually makes a big difference.

This change implements the plan outlined by @simonpj in
https://mail.haskell.org/pipermail/ghc-devs/2017-October/014974.html
which is basically about using a map and constructing it outside the
loop.

One place where things got a bit awkward was TysWiredIn.hs,
it would have been possible to just assign the tags by hand, but
that seemed error-prone to me, so I decided to go through a map
there as well.

Test Plan:
./validate
On a file with 10k constructors
Before:
   8,130,522,344 bytes allocated in the heap
  Total   time    3.682s  (  3.920s elapsed)
After:
   4,133,478,744 bytes allocated in the heap
  Total   time    2.509s  (  2.750s elapsed)

Reviewers: simonpj, bgamari

Reviewed By: simonpj

Subscribers: goldfire, rwbarton, thomie, simonmar, carter, simonpj

GHC Trac Issues: #14657

Differential Revision: https://phabricator.haskell.org/D4289

These were shown up by Trac #14643

Bug 1: if we had a single partial signature for
two functions
   f, g :: forall a. _ -> a
then we made two different SigTvs but with the sane Name.
This was jolly confusing and ultimately led to deeply bogus
results with Any's appearing in the resulting program. Yikes.
Fix: clone the quantified variables in TcSigs.tcInstSig (as
indeed its name suggests).

Bug 2: we were not eliminating duplicate/superclass constraints
in the partial signatures of a mutually recursive group.

Easy to fix: we are already doing dup/superclass elim in
TcSimplify.decideQuantification.  So we move the partial-sig
constraints there too.

This patch moves the "ok_unfolding" test
   from  CoreOpt.joinPointBinding_maybe
   to    OccurAnal.decideJoinPointHood

Previously the occurrence analyser was deciding to make
something a join point, but the simplifier was reversing
that decision, which made the decision about /other/ bindings
invalid.

Fixes Trac #14650.

Summary:
get/setAllocationCounter didn't take into account allocations in the
current block. This was known at the time, but it turns out to be
important to have more accuracy when using these in a fine-grained
way.

Test Plan:
New unit test to test incrementally larger allocaitons.  Before I got
results like this:

```
+0
+0
+0
+0
+0
+4096
+0
+0
+0
+0
+0
+4064
+0
+0
+4088
+4056
+0
+0
+0
+4088
+4096
+4056
+4096
```

Notice how the results aren't always monotonically increasing.  After
this patch:

```
+344
+416
+488
+560
+632
+704
+776
+848
+920
+992
+1064
+1136
+1208
+1280
+1352
+1424
+1496
+1568
+1640
+1712
+1784
+1856
+1928
+2000
+2072
+2144
```

Reviewers: niteria, bgamari, hvr, erikd

Subscribers: rwbarton, thomie, carter

Differential Revision: https://phabricator.haskell.org/D4288

This one, discovered by Iceland Jack (Trac #14507), shows
that a pattern-bound coercion can show up in the argument
type(s) of the matcher of a pattern synonym.

The error message isn't great, but at least we now rightly
reject the program.

Just better layout in output for the user

Trac #14552 showed a very obscure case where we can't infer a
good pattern-synonym type.

The error message is horrible, but at least we no longer crash
and burn.

Trac #13032 pointed out that we sometimes generate unused
bindings for Givens, and (worse still) we can't always discard
them later (we don't drop a case binding unless we can prove
that the scrutinee is non-bottom.

It looks as if this may be a major reason for the performace
problems in #14338 (see comment:29).

This patch fixes the problem at source, by pruning away all the
dead Givens.  See Note [Delete dead Given evidence bindings]

Remarkably, compiler allocation falls by 23% in
perf/compiler/T12227!

I have not confirmed whether this change actualy helps with