[Git][ghc/ghc][wip/pmcheck-ncon] 53 commits: TestRunner: Added --chart to display a chart of performance tests

Sebastian Graf gitlab at gitlab.haskell.org
Mon Jun 10 20:52:30 UTC 2019 


Sebastian Graf pushed to branch wip/pmcheck-ncon at Glasgow Haskell Compiler / GHC


Commits:
286827be by David Eichmann at 2019-06-04T05:09:05Z
TestRunner: Added --chart to display a chart of performance tests

This uses the Chart.js javascript library.
Everything is put into a standalone .html file and opened with the
default browser.
I also simplified the text output to use the same data as the chart.
You can now use a commit range with git's ".." syntax.
The --ci option will use results from CI (you'll need to fetch them
first):

  $ git fetch https://gitlab.haskell.org/ghc/ghc-performance-notes.git refs/notes/perf:refs/notes/ci/perf
  $ python3 testsuite/driver/perf_notes.py --ci --chart --test-env x86_64-darwin --test-name T9630 master~500..master

- - - - -
db78ac6f by Andrew Martin at 2019-06-04T05:09:43Z
Use a better strategy for determining the offset applied to foreign function arguments that have an unlifted boxed type. We used to use the type of the argument. We now use the type of the foreign function. Add a test to confirm that the roundtrip conversion between an unlifted boxed type and Any is sound in the presence of a foreign function call.

- - - - -
114b014f by Alp Mestanogullari at 2019-06-04T05:10:20Z
Hadrian: fix OSX build failure and add an OSX/Hadrian CI job

The OSX build failure introduced in 3aa71a22 was due to a change in the
glob we use to collect libffi shared libraries in hadrian/src/Rules/Libffi.hs.
This commit fixes the problem and adds an OSX CI job that builds GHC with
Hadrian, to make sure we don't break it again.

- - - - -
002594b7 by Xavier Denis at 2019-06-04T18:41:29Z
Add GHCi :instances command

This commit adds the `:instances` command to ghci following proosal
number 41.

This makes it possible to query which instances are available to a given
type.

The output of this command is all the possible instances with type
variables and constraints instantiated.

- - - - -
3ecc03df by Ben Gamari at 2019-06-04T18:42:04Z
gitlab-ci: Run bindisttest during CI

- - - - -
c16f3297 by Ben Gamari at 2019-06-04T18:42:04Z
make: Fix bindist installation

This fixes a few vestigial references to `settings` left over from !655.
Fixes #16715.

- - - - -
ba4e3934 by Alp Mestanogullari at 2019-06-04T18:43:17Z
Hadrian: profiling and debug enabled ways support -eventlog too

- - - - -
567894b4 by Matthew Pickering at 2019-06-07T08:36:32Z
gitlab-ci: Disable darwin hadrian job

See #16771

We don't have enough capacity for the two jobs currently.

[skip ci]

- - - - -
d3915b30 by Andrew Martin at 2019-06-07T14:20:42Z
[skip ci] Improve the documentation of the CNF primops. In this context, the term "size" is ambiguous and is now avoided. Additionally, the distinction between a CNF and the blocks that comprise it has been emphasize. The vocabulary has been made more consistent with the vocabulary in the C source for CNF.

- - - - -
e963beb5 by Sebastian Graf at 2019-06-07T14:21:21Z
TmOracle: Replace negative term equalities by refutable PmAltCons

The `PmExprEq` business was a huge hack and was at the same time vastly
too powerful and not powerful enough to encode negative term equalities,
i.e. facts of the form "forall y. x ≁ Just y".

This patch introduces the concept of 'refutable shapes': What matters
for the pattern match checker is being able to encode knowledge of the
kind "x can no longer be the literal 5". We encode this knowledge in a
`PmRefutEnv`, mapping a set of newly introduced `PmAltCon`s (which are
just `PmLit`s at the moment) to each variable denoting above
inequalities.

So, say we have `x ≁ 42 ∈ refuts` in the term oracle context and
try to solve an equality like `x ~ 42`. The entry in the refutable
environment will immediately lead to a contradiction.

This machinery renders the whole `PmExprEq` and `ComplexEq` business
unnecessary, getting rid of a lot of (mostly dead) code.

See the Note [Refutable shapes] in TmOracle for a place to start.

Metric Decrease:
    T11195

- - - - -
0b7372f6 by Matthew Pickering at 2019-06-07T14:21:57Z
Add HEAP_PROF_SAMPLE_END event to mark end of samples

This allows a user to observe how long a sampling period lasts so that
the time taken can be removed from the profiling output.

Fixes #16697

- - - - -
d1dc0ed7 by Roland Senn at 2019-06-07T14:22:47Z
Fix #16700: Tiny errors in output of GHCi commands :forward and :info

`:info Coercible` now outputs the correct section number of the GHCi User's guide together with the secion title.

`:forward x` gives the correct syntax hint.

- - - - -
387050d0 by John Ericson at 2019-06-07T14:23:23Z
Factor out 'getLibDir' / 'getBaseDir' into a new GHC.BaseDir ghc-boot module

ghc-pkg and ghc already both needed this. I figure it is better to
deduplicate, especially seeing that changes to one (FreeBSD CPP) didn't
make it to the other.

Additionally in !1090 I make ghc-pkg look up the settings file, which
makes it use the top dir a bit more widely. If that lands, any
difference in the way they find the top dir would be more noticable.

That change also means sharing more code between ghc and ghc-package
(namely the settings file parsing code), so I'd think it better to get
off the slipperly slope of duplicating code now.

- - - - -
da26ffe7 by Simon Peyton Jones at 2019-06-07T14:24:00Z
Preserve ShadowInfo when rewriting evidence

When the canonicaliser rewrites evidence of a Wanted, it
should preserve the ShadowInfo (ctev_nosh) field.  That is,
a WDerive should rewrite to WDerive, and WOnly to WOnly.

Previously we were unconditionally making a WDeriv, thereby
rewriting WOnly to WDeriv.  This bit Nick Frisby (issue #16735)
in the context of his plugin, but we don't have a compact test
case.

The fix is simple, but does involve a bit more plumbing,
to pass the old ShadowInfo around, to use when building
the new Wanted.

- - - - -
9bb58799 by Ben Gamari at 2019-06-07T14:24:38Z
Hadrian: Delete target symlink in createFileLinkUntracked

Previously createFileLinkUntracked would fail if the symlink already
existed.

- - - - -
be63d299 by Simon Jakobi at 2019-06-07T14:25:16Z
Fix isValidNatural: The BigNat in NatJ# must have at least 2 limbs

Previously the `integer-gmp` variant of `isValidNatural` would fail to
detect values `<= maxBound::Word` that were incorrectly encoded using
the `NatJ#` constructor.

- - - - -
e87b9f87 by Moritz Angermann at 2019-06-07T14:26:04Z
llvm-targets: Add x86_64 android layout

- - - - -
60db142b by code5hot at 2019-06-07T14:26:46Z
Update Traversable.hs with a note about an intuitive law
- - - - -
f11aca52 by code5hot at 2019-06-07T14:26:46Z
Used terminology from a paper. Added it as a reference.
- - - - -
13b3d45d by code5hot at 2019-06-07T14:26:46Z
remove backticks from markup - it doesn't mean what I think it means
- - - - -
cfd3e0f1 by Zejun Wu at 2019-06-07T14:27:34Z
Pass preprocessor options to C compiler when building foreign C files (#16737)

- - - - -
5991d877 by Ben Gamari at 2019-06-07T14:28:09Z
base: Export Finalizers

As requested in #16750.

- - - - -
3d97bad6 by Alp Mestanogullari at 2019-06-07T14:28:47Z
Hadrian: use deb9 Docker images instead of deb8 for CI jobs

This should fix #16739, where we seem to be getting extra carets in
a test's output because of the gcc that ships with the deb8 image,
whule we're not observing those extra carets in the deb9-based (Make)
jobs.

- - - - -
1afb4995 by Ben Gamari at 2019-06-07T14:29:23Z
gitlab-ci: Create index.html in documentation deployment

Otherwise navigating to https://ghc.gitlab.haskell.org/ghc will result
in a 404.

- - - - -
07dc79c3 by Matthew Pickering at 2019-06-08T17:34:18Z
gitlab-ci: Linters, don't allow to fail

Ben disabled them in cd85f8a71bb56cff332560e1d571b3406789fb71 but didn't
say how or why they were broken.

- - - - -
fd840b64 by Matthew Pickering at 2019-06-08T17:34:18Z
gitlab-ci: Don't run two submodule checking jobs on Marge jobs

- - - - -
310d0c4c by Matthew Pickering at 2019-06-08T17:34:18Z
Fix two lint failures in rts/linker/MachO.c

- - - - -
fe965316 by Ben Gamari at 2019-06-08T17:34:18Z
gitlab-ci: Use --unshallow when fetching for linters

GitLab creates a shallow clone. However, this means that we may not have
the base commit of an MR when linting, causing `git merge-base` to fail.
Fix this by passing `--unshallow` to `git fetch`, ensuring that we have
the entire history.

- - - - -
f58234ea by Ben Gamari at 2019-06-08T17:34:18Z
gitlab-ci: Fix submodule linter

The job script didn't even try to compute the base commit to lint with
respect to.

- - - - -
c392f987 by Ben Gamari at 2019-06-08T17:34:18Z
gitlab-ci: A few clarifying comments

- - - - -
709290b0 by Matthew Pickering at 2019-06-08T17:38:15Z
Remove trailing whitespace

[skip ci]

This should really be caught by the linters! (#16711)

- - - - -
b2f106f5 by Ben Gamari at 2019-06-08T18:02:02Z
gitlab-ci: Disable shallow clones

Previously we were passing `--unshallow` to `git fetch` in the linting
rules to ensure that the base commit which we were linting with respect
to was available. However, this breaks due to GitLab's re-use of
working directories since `git fetch --unshallow` fails on a repository
which is not currently shallow.

Given that `git fetch --unshallow` circumvents the efficiencies provided
by shallow clones anyways, let's just disable them entirely.

There is no documented way to do disable shallow clones but on checking
the GitLab implementation it seems that setting `GIT_DEPTH=0` should do
the trick.

- - - - -
4a72259d by Ben Gamari at 2019-06-08T18:40:55Z
gitlab-ci: Fix submodule linting of commits

There is no notion of a base commit when we aren't checking a merge
request. Just check the HEAD commit.

- - - - -
87540029 by Ben Gamari at 2019-06-08T20:44:55Z
gitlab-ci: Ensure that all commits on a branch are submodule-linted

The previous commit reworked things such that the submodule linter would
only run on the head commit. However, the linter only checks the
submodules which are touched by the commits it is asked to lint.
Consequently it would be possible for a bad submodule to sneak through.

Thankfully, we can use the handy CI_COMMIT_BEFORE_SHA attribute to
find the base commit of the push.

- - - - -
0462b0e0 by Alexandre Baldé at 2019-06-09T15:48:34Z
Explain that 'mappend' and '(<>)' should be the same [skip ci]

- - - - -
970e4802 by Matthew Pickering at 2019-06-09T15:49:09Z
hadrian: Properly partition options in sourceArgs

Previously if you build the `ghc` package then it would has the default
opts and the library opts. This is different behaviour to make where the
library opts are only reserved for things in the `libraries`
subdirectory (I believe)

Fixes #16716

- - - - -
a018c3a8 by Ben Gamari at 2019-06-09T15:49:44Z
testsuite: Suppress ticks in T4918 output

As noted in #16741, this test otherwise breaks when `base` is compiled
with `-g`.

- - - - -
f7370333 by chessai at 2019-06-09T22:41:02Z
Introduce log1p and expm1 primops

Previously log and exp were primitives yet log1p and expm1 were FFI
calls. Fix this non-uniformity.

- - - - -
41bf4045 by Ben Gamari at 2019-06-09T22:41:38Z
testsuite: Add test for #16514

- - - - -
b9fe91fc by Simon Jakobi at 2019-06-09T22:42:21Z
Small refactorings in ExtractDocs

- - - - -
9d238791 by Kevin Buhr at 2019-06-09T22:42:57Z
Handle trailing path separator in package DB names (#16360)

Package DB directories with trailing separator (provided via
GHC_PACKAGE_PATH or via -package-db) resulted in incorrect calculation of
${pkgroot} substitution variable.  Keep the trailing separator while
resolving as directory or file, but remove it before dropping the last
path component with takeDirectory.

Closes #16360.

- - - - -
a22e51ea by Richard Eisenberg at 2019-06-09T22:43:38Z
Fix #16517 by bumping the TcLevel for method sigs

There were actually two bugs fixed here:

1. candidateQTyVarsOfType needs to be careful that it does not
   try to zap metavariables from an outer scope as "naughty"
   quantification candidates. This commit adds a simple check
   to avoid doing so.

2. We weren't bumping the TcLevel in kcHsKindSig, which was used
   only for class method sigs. This mistake led to the acceptance
   of

     class C a where
       meth :: forall k. Proxy (a :: k) -> ()

   Note that k is *locally* quantified. This patch fixes the
   problem by using tcClassSigType, which correctly bumps the
   level. It's a bit inefficient because tcClassSigType does other
   work, too, but it would be tedious to repeat much of the code
   there with only a few changes. This version works well and is
   simple.

And, while updating comments, etc., I noticed that tcRnType was
missing a pushTcLevel, leading to #16767, which this patch also
fixes, by bumping the level. In the refactoring here, I also
use solveEqualities. This initially failed ghci/scripts/T15415,
but that was fixed by teaching solveEqualities to respect
-XPartialTypeSignatures.

This patch also cleans up some Notes around error generation that
came up in conversation.

Test case: typecheck/should_fail/T16517, ghci/scripts/T16767

- - - - -
10452959 by Roland Senn at 2019-06-09T22:44:18Z
Add disable/enable commands to ghci debugger #2215

This patch adds two new commands `:enable` and `:disable` to the GHCi debugger.
Opposite to `:set stop <n> :continue` a breakpoint disabled with `:disable` will
not loose its previously set stop command.
A new field breakEnabled is added to the BreakLocation data structure to
track the enable/disable state. When a breakpoint is disabled with a `:disable`
command, the following happens:

The corresponding BreakLocation data element is searched dictionary of the
`breaks` field of the GHCiStateMonad. If the break point is found and not
already in the disabled state, the breakpoint is removed from bytecode.
The BreakLocation data structure is kept in the breaks list and the new
breakEnabled field is set to false.

The `:enable` command works similar.

The breaks field in the GHCiStateMonad was changed from an association list
to int `IntMap`.

- - - - -
13572480 by Ben Gamari at 2019-06-09T22:44:54Z
rts: Separate population of eventTypes from initial event generation

Previously these two orthogonal concerns were both implemented in
postHeaderEvents which made it difficult to send header events after RTS
initialization.

- - - - -
ed20412a by nineonine at 2019-06-09T22:45:31Z
Do not report error if Name in pragma is unbound

- - - - -
8a48a8a4 by Ben Gamari at 2019-06-09T22:46:08Z
testsuite: Add test for #16509

- - - - -
69c58f8a by David Eichmann at 2019-06-09T22:46:46Z
Hadrian: need CPP preprocessor dependencies #16660

Use the new -include-cpp-deps ghc option (#16521)
when generating .dependencies files in hadrian.
This is version gated as -include-cpp-deps is a
relatively new option.

- - - - -
1c7bb03d by Richard Eisenberg at 2019-06-09T22:47:24Z
Comments only: document tcdDataCusk better.

- - - - -
5023adce by John Ericson at 2019-06-09T22:47:59Z
Remove CPP ensuring word size is 32 or 64 bits around Addr# <-> int# primops

It shouldn't be needed these days, and those primops are "highly
deprecated" anyways.

This fits with my plans because it removes one bit of target-dependence
of the builtin primops, and this is the hardest part of GHC to make
multi-target.

CC @carter

- - - - -
8e60e3f0 by Daniel Gröber at 2019-06-09T22:48:38Z
rts: Fix RetainerProfile early return with TREC_CHUNK

When pop() returns with `*c == NULL` retainerProfile will immediately
return. All other code paths is pop() continue with the next stackElement
when this happens so it seems weird to me that TREC_CHUNK we would suddenly
abort everything even though the stack might still have elements left to
process.

- - - - -
1a3420ca by Ben Gamari at 2019-06-10T11:59:41Z
base: Mark CPUTime001 as fragile

As noted in #16224, CPUTime001 has been quite problematic, reporting
non-monotonic timestamps in CI. Unfortunately I've been unable to
reproduce this locally.

- - - - -
9bc10993 by Vladislav Zavialov at 2019-06-10T12:00:16Z
Print role annotations in TemplateHaskell brackets (#16718)

- - - - -
8282026b by Sebastian Graf at 2019-06-10T20:51:52Z
Add `PmNCons` to `Check` for correct warnings in the presence of `COMPLETE` groups

Previously, we had an elaborate mechanism for selecting the warnings to
generate in the presence of different `COMPLETE` matching groups that,
albeit finely-tuned, produced wrong results from an end user's
perspective in some cases (#13363).

The underlying issue is that at the point where the `ConVar` case has to
commit to a particular `COMPLETE` group, there's not enough information
to do so and the status quo was to just enumerate all possible complete
sets nondeterministically.
The `getResult` function would then pick the outcome according to
metrics defined in accordance to the user's guide. But crucially, it
lacked knowledge about the order in which affected clauses appear,
leading to the surprising behavior in #13363.

The introduction of an `PmNCons` variant in `PmPat` fixes this: Instead
of committing to a particular `COMPLETE` group in the `ConVar` case,
we now split off the matching constructor incrementally and record the
newly covered case in `PmNCons`.
After all clauses have been processed this way, we filter out any value
vector abstractions from the uncovered set involving `PmNCons` whose set
of covered constructors completely overlap a `COMPLETE` set.

- - - - -


27 changed files:

- .gitlab-ci.yml
- compiler/basicTypes/NameEnv.hs
- compiler/cmm/CmmMachOp.hs
- compiler/cmm/MkGraph.hs
- compiler/cmm/PprC.hs
- compiler/codeGen/StgCmmExpr.hs
- compiler/codeGen/StgCmmForeign.hs
- compiler/codeGen/StgCmmPrim.hs
- compiler/deSugar/Check.hs
- compiler/deSugar/DsMonad.hs
- compiler/deSugar/ExtractDocs.hs
- compiler/deSugar/PmExpr.hs
- + compiler/deSugar/PmPpr.hs
- compiler/deSugar/TmOracle.hs
- compiler/ghc.cabal.in
- compiler/ghci/LinkerTypes.hs
- compiler/hsSyn/HsDecls.hs
- compiler/llvmGen/LlvmCodeGen/CodeGen.hs
- compiler/main/DriverPipeline.hs
- compiler/main/GHC.hs
- compiler/main/HscMain.hs
- compiler/main/HscTypes.hs
- compiler/main/InteractiveEval.hs
- compiler/main/Packages.hs
- compiler/main/SysTools/BaseDir.hs
- compiler/nativeGen/PPC/CodeGen.hs
- compiler/nativeGen/SPARC/CodeGen.hs


The diff was not included because it is too large.


View it on GitLab: https://gitlab.haskell.org/ghc/ghc/compare/17739b9aaf03dc90d4cf961c367ea47156c5bede...8282026bd37270e50344024c1c705b5b2b3ef98e

-- 
View it on GitLab: https://gitlab.haskell.org/ghc/ghc/compare/17739b9aaf03dc90d4cf961c367ea47156c5bede...8282026bd37270e50344024c1c705b5b2b3ef98e
You're receiving this email because of your account on gitlab.haskell.org.


-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.haskell.org/pipermail/ghc-commits/attachments/20190610/d08fd3ea/attachment-0001.html>

More information about the ghc-commits mailing list