bgamari (Ben Gamari)Administrator
Blub... blub... blub...

Today

  • Clear sailing ahead.

Tomorrow

  • Clear sailing ahead.

Friday

  • Clear sailing ahead.

Badges

User Since
Jul 20 2014, 6:05 AM (135 w, 3 d)
Roles
Administrator
Availability
Available

I like working on GHC. 'tis all I have to say.

Recent Activity

Yesterday

bgamari committed rGHC8ccbc2e5252a: Bump Cabal and containers submodules (authored by bgamari).
Bump Cabal and containers submodules
Tue, Feb 21, 6:42 PM
bgamari committed rGHC9ef2749c3f38: Fix all broken perf tests on x64 Windows (authored by Phyx).
Fix all broken perf tests on x64 Windows
Tue, Feb 21, 6:42 PM
bgamari closed D3160: Fix all broken perf tests on x64 Windows by committing rGHC9ef2749c3f38: Fix all broken perf tests on x64 Windows (authored by Phyx).
Tue, Feb 21, 6:42 PM
bgamari accepted D3169: Prohibit RULES changing constructors.

Looks quite good! Just point small suggestions.

Tue, Feb 21, 5:08 PM
bgamari requested changes to D3169: Prohibit RULES changing constructors.

As you say, documentation is needed here:

  • a mention in the release notes
  • a mention in the users guide in the discussion of the RULES pragma
Tue, Feb 21, 3:22 PM
bgamari updated the Trac tickets for D3169: Prohibit RULES changing constructors.
Tue, Feb 21, 3:20 PM
bgamari accepted D3156: real/Makefile: remove the mention of ebnf2ps.
Tue, Feb 21, 2:41 PM
bgamari accepted D3157: real: enable linear.
Tue, Feb 21, 2:41 PM
bgamari accepted D3159: real: remove `rx`.
Tue, Feb 21, 2:40 PM
bgamari accepted D3161: spectral: enable mate.

Thanks!

Tue, Feb 21, 2:40 PM
bgamari added a comment to D3132: [RFC] JSON profiler reports.
In D3132#92469, @maoe wrote:

See https://gist.github.com/bgamari/699c24a83af6f7d61c5e2fa4b96e17c4 for one such example.

Thank you. It seems that the JSON file is malformed. There is no closing bracket for "cost_centers".

Tue, Feb 21, 2:39 PM
bgamari added a comment to D3132: [RFC] JSON profiler reports.
In D3132#91992, @maoe wrote:

For the record, the output of this looks something like,

...

It'd be great if you could attach a full JSON output or two to experiment the tools. I think there should be some lines including "profile": { between the end of "cost_centres" and the first "children: [.

Tue, Feb 21, 11:49 AM
bgamari accepted D3160: Fix all broken perf tests on x64 Windows.
Tue, Feb 21, 10:31 AM
bgamari accepted D3136: Add -fwhole-archive-hs-libs.

Alright, I think this looks fine despite the rather unfortunate name.

Tue, Feb 21, 10:06 AM
bgamari accepted D3154: Make list of deprecated symbols on Windows weak..

Yes, this seems reasonable.

Tue, Feb 21, 10:03 AM
bgamari accepted D3155: Load `pthreads` by default on Windows.

It's a bit yucky, but I suppose fine for now.

Tue, Feb 21, 10:01 AM
bgamari accepted D3158: Correct Windows libdir assumptions..

Looks reasonable to me.

Tue, Feb 21, 9:59 AM
bgamari added a comment to D3164: Use half as much memory when reading interfaces.

@simonmar reports on the audit that he has no idea why the original (*2) was added. In light of this fact, and that this appears not to break anything, let's ship it.

Tue, Feb 21, 9:53 AM
bgamari added a comment to rGHC27a2854124cc: A number of Typeable wibbles from review.
In null, @nomeata wrote:

Runtime performance improvements in fasta and lambda:
https://perf.haskell.org/ghc/#revision/27a2854124cc1c101570104501beea234a4ee921
A bit weird, isn’t it?

But it clearly fixes the regression in https://perf.haskell.org/ghc/#revision/8fa4bf9ab3f4ea4b208f4a43cc90857987e6d497

Tue, Feb 21, 9:51 AM
bgamari accepted D3028: Load dependent dlls..

Alright, looks good to me. Good catch regarding the multiple loading issue, @rwbarton.

Tue, Feb 21, 9:49 AM
bgamari accepted D3137: Set $1_$2_SplitSections in distdir-opts.mk not build-package.mk.

Alright, sounds good to me.

Tue, Feb 21, 9:31 AM
bgamari accepted D3165: Changelog notice for compact..

Thanks @ezyang!

Tue, Feb 21, 9:07 AM
bgamari committed rGHC6e3288473718: Fix SetLevels for join points (authored by simonpj).
Fix SetLevels for join points
Tue, Feb 21, 8:31 AM
bgamari committed rGHCe790126cd57a: Improve Core Lint, mainly for join points (authored by simonpj).
Improve Core Lint, mainly for join points
Tue, Feb 21, 8:31 AM
bgamari committed rGHC82694e6765da: testsuite: Fix allocations of T10547 (authored by bgamari).
testsuite: Fix allocations of T10547
Tue, Feb 21, 8:31 AM

Mon, Feb 20

bgamari updated the diff for D3166: TcTypeable: Try to reuse KindReps.

Fix test output

Mon, Feb 20, 9:43 PM
bgamari created D3166: TcTypeable: Try to reuse KindReps.
Mon, Feb 20, 8:41 PM
bgamari committed rGHCb7265ffa06d5: build.mk: Add option for debug symbols (authored by bgamari).
build.mk: Add option for debug symbols
Mon, Feb 20, 6:20 PM
bgamari committed rGHC7153370288e6: Bump time submodule to 1.8 (authored by bgamari).
Bump time submodule to 1.8
Mon, Feb 20, 6:20 PM
bgamari committed rGHC240b43e49695: Bump Win32 submodule to 2.5.1.0 (authored by bgamari).
Bump Win32 submodule to 2.5.1.0
Mon, Feb 20, 6:20 PM
bgamari accepted D3147: Fix #13287 adding handling of -- to getArgs for Windows..

Thanks @AndreasK!

Mon, Feb 20, 4:21 PM
bgamari added a comment to D3158: Correct Windows libdir assumptions..

@Phyx, is it possible you chose the wrong base commit while updating this diff?

Mon, Feb 20, 2:58 PM
bgamari requested changes to D3145: Add HsEmbellished type to hsSyn.

@alanz, we discussed this on the GHC call and the general consensus was that this approach seems unnecessarily expensive. Why not instead represent these embellished names with new constructors of HsExpr, following the example of HsPar. Yes, this does allow us to represent invalid expressions like,

a `hello 5` 8

but I doubt this is the first place where HsExpr allows representation of invalid programs. Is there a reason this approach won't work?

Mon, Feb 20, 2:02 PM
bgamari updated subscribers of rGHC526c3af1dc98: Use MD5 checksums for recompilation checking (fixes #1372, #1959).
Mon, Feb 20, 1:58 PM
bgamari accepted D3164: Use half as much memory when reading interfaces.

The original (*2) is quite odd. Regardless, this looks fine to me.

Mon, Feb 20, 1:53 PM
bgamari added a comment to D3160: Fix all broken perf tests on x64 Windows.

Yay, thanks for taking care of this @Phyx! I can rebase this.

Mon, Feb 20, 1:52 PM
bgamari committed rGHC6ad89d7bacc9: Bump a few more performance regressions from Type-indexed Typeable (authored by bgamari).
Bump a few more performance regressions from Type-indexed Typeable
Mon, Feb 20, 12:42 PM
bgamari committed rGHC087dbbedc4ca: Bump Cabal submodule (authored by bgamari).
Bump Cabal submodule
Mon, Feb 20, 12:42 PM
bgamari committed rGHC27a2854124cc: A number of Typeable wibbles from review (authored by bgamari).
A number of Typeable wibbles from review
Mon, Feb 20, 12:42 PM
bgamari committed rGHC8dd4e3bbbaa3: Remove redundant import (authored by simonpj).
Remove redundant import
Mon, Feb 20, 12:42 PM
bgamari committed rGHC3f653c1ec62e: Fix Core pretty printer (authored by simonpj).
Fix Core pretty printer
Mon, Feb 20, 12:42 PM
bgamari committed rGHC8a9b57f6c16e: Kill off the remaining Rec [] (authored by simonpj).
Kill off the remaining Rec []
Mon, Feb 20, 12:42 PM
bgamari committed rGHC2d5be63d1140: Change -dppr-ticks to -dsuppress-ticks (authored by simonpj).
Change -dppr-ticks to -dsuppress-ticks
Mon, Feb 20, 12:42 PM

Sat, Feb 18

bgamari added a comment to D3061: Remove unused PostTc types from TuplePat.

Ping.

Sat, Feb 18, 8:56 PM
bgamari requested changes to D3113: Eliminate ListSetOps from imp_trust_pkgs.

There still appear to be build issues here.

Sat, Feb 18, 8:56 PM
bgamari accepted D3140: testsuite: Remove old python version tests.

Yay!

Sat, Feb 18, 8:54 PM
bgamari abandoned D2010: Type-indexed type representations.

This was merged in 8fa4bf9ab3f4ea4b208f4a43cc90857987e6d497

Sat, Feb 18, 8:54 PM
bgamari abandoned D2038: TysPrim: Generalize kind of (->).

This was merged in b207b536ded40156f9adb168565ca78e1eef2c74.

Sat, Feb 18, 8:53 PM
bgamari committed rNOFIB3ac6f2db0925: real: remove HMMS (authored by michalt).
real: remove HMMS
Sat, Feb 18, 1:24 AM
bgamari closed D3088: real: remove HMMS by committing rNOFIB3ac6f2db0925: real: remove HMMS (authored by michalt).
Sat, Feb 18, 1:24 AM
bgamari committed rNOFIB8a11a3859bb6: real: remove ebnf2ps (authored by michalt).
real: remove ebnf2ps
Sat, Feb 18, 1:24 AM
bgamari closed D3087: real: remove ebnf2ps by committing rNOFIB8a11a3859bb6: real: remove ebnf2ps (authored by michalt).
Sat, Feb 18, 1:24 AM
bgamari committed rNOFIB5786292b9473: spectral: revive lambda (authored by michalt).
spectral: revive lambda
Sat, Feb 18, 1:24 AM
bgamari closed D3079: spectral: revive lambda by committing rNOFIB5786292b9473: spectral: revive lambda (authored by michalt).
Sat, Feb 18, 1:24 AM
bgamari committed rNOFIB0a120575afb6: imaginary: re-enable digits-of-e{1,2} (authored by michalt).
imaginary: re-enable digits-of-e{1,2}
Sat, Feb 18, 1:24 AM
bgamari closed D3078: imaginary: re-enable digits-of-e{1,2} by committing rNOFIB0a120575afb6: imaginary: re-enable digits-of-e{1,2} (authored by michalt).
Sat, Feb 18, 1:24 AM
bgamari committed rNOFIB070e01659b53: spectral: revive last-piece (authored by michalt).
spectral: revive last-piece
Sat, Feb 18, 1:24 AM
bgamari closed D3042: spectral: revive last-piece by committing rNOFIB070e01659b53: spectral: revive last-piece (authored by michalt).
Sat, Feb 18, 1:24 AM
bgamari committed rNOFIBdf44189dd7b9: nucleic2: removed unused files & simplify Makefile (authored by michalt).
nucleic2: removed unused files & simplify Makefile
Sat, Feb 18, 1:24 AM
bgamari closed D3041: nucleic2: removed unused files & simplify Makefile by committing rNOFIBdf44189dd7b9: nucleic2: removed unused files & simplify Makefile (authored by michalt).
Sat, Feb 18, 1:24 AM
bgamari committed rNOFIB307ae61121a9: Fix nofib benchmarks on Windows. (authored by Phyx).
Fix nofib benchmarks on Windows.
Sat, Feb 18, 1:24 AM
bgamari closed D3030: Fix nofib benchmarks on Windows. by committing rNOFIB307ae61121a9: Fix nofib benchmarks on Windows. (authored by Phyx).
Sat, Feb 18, 1:24 AM

Fri, Feb 17

bgamari accepted D3148: Improve Haddock documentation for compact..

I left one point of improvement inline, but otherwise good. Thanks!

Fri, Feb 17, 11:24 PM
bgamari accepted D3150: ghci users guide: mention "~" expansion in :add.

Thanks Chris!

Fri, Feb 17, 11:12 PM
bgamari committed rGHC42ff5d97b486: Disable Typeable binding generation for unboxed sums (authored by bgamari).
Disable Typeable binding generation for unboxed sums
Fri, Feb 17, 11:11 PM
bgamari committed rGHC8fa4bf9ab3f4: Type-indexed Typeable (authored by bgamari).
Type-indexed Typeable
Fri, Feb 17, 11:11 PM
bgamari committed rGHCb207b536ded4: Generalize kind of the (->) tycon (authored by bgamari).
Generalize kind of the (->) tycon
Fri, Feb 17, 11:11 PM
bgamari committed rGHCefeaf9e43610: Bump nofib submodule (authored by bgamari).
Bump nofib submodule
Fri, Feb 17, 11:11 PM
bgamari committed rGHC7c057b50e7c2: Bump libraries/array submodule (authored by bgamari).
Bump libraries/array submodule
Fri, Feb 17, 11:10 PM
bgamari updated the diff for D3149: configure: detect whether -lpthreads is necessary for pthreads.

Fix parens

Fri, Feb 17, 9:17 PM
bgamari created D3149: configure: detect whether -lpthreads is necessary for pthreads.
Fri, Feb 17, 9:11 PM
bgamari retitled D3147: Fix #13287 adding handling of -- to getArgs for Windows. from Summary: Fix #13287 adding handling of -- to getArgs for Windows. to Fix #13287 adding handling of -- to getArgs for Windows..
Fri, Feb 17, 4:31 PM
bgamari abandoned D3142: Track good joins instead of bad.

This was addressed in 6bab649bde653f13c15eba30d5007bef4a9a9d3a.

Fri, Feb 17, 4:31 PM
bgamari requested verification of rGHC3eb737ee3f90: Generalize CmmUnwind and pass unwind information through NCG.

I believe this is now resolved.

Fri, Feb 17, 4:30 PM

Wed, Feb 15

bgamari added a comment to D3132: [RFC] JSON profiler reports.

For the record, I also have some hacked together analysis tools for this format.

Wed, Feb 15, 9:54 AM

Tue, Feb 14

bgamari requested changes to D2929: Always expose unfoldings for overloaded functions.

Bumping out of the review queue for now.

Tue, Feb 14, 4:27 PM
bgamari requested changes to D2996: WIP – computed gotos in the interpreter.

Bumping out of the review queue for now. @DemiMarie, let me know if you need help.

Tue, Feb 14, 4:26 PM
bgamari requested changes to D3130: Fix a Backpack recompilation avoidance bug when signatures change..
Tue, Feb 14, 4:26 PM
bgamari added a comment to D3130: Fix a Backpack recompilation avoidance bug when signatures change..

This generally looks quite good save a missing comment and build issues.

Tue, Feb 14, 4:26 PM
bgamari added a comment to D3136: Add -fwhole-archive-hs-libs.

Also, we should probably have a test that links an executable using this flag. Perhaps we could check for the presence of the symbol using strings.

Tue, Feb 14, 4:16 PM
bgamari requested changes to D3136: Add -fwhole-archive-hs-libs.

Requesting changes due to OS X issue.

Tue, Feb 14, 4:15 PM
bgamari committed rGHC93e65c8ab7b4: Don't warn about missing methods for instances in signatures. (authored by ezyang).
Don't warn about missing methods for instances in signatures.
Tue, Feb 14, 3:58 PM
bgamari closed D3134: Don't warn about missing methods for instances in signatures. by committing rGHC93e65c8ab7b4: Don't warn about missing methods for instances in signatures. (authored by ezyang).
Tue, Feb 14, 3:58 PM
bgamari committed rGHCf90e61ad6e5f: Make deSugarExpr use runTcInteractive (authored by rwbarton).
Make deSugarExpr use runTcInteractive
Tue, Feb 14, 3:58 PM
bgamari closed D3125: Make deSugarExpr use runTcInteractive by committing rGHCf90e61ad6e5f: Make deSugarExpr use runTcInteractive (authored by rwbarton).
Tue, Feb 14, 3:58 PM
bgamari committed rGHCbedcb7165925: Check local type family instances against all imported ones (authored by rwbarton).
Check local type family instances against all imported ones
Tue, Feb 14, 3:58 PM
bgamari closed D2992: Check local type family instances against all imported ones by committing rGHCbedcb7165925: Check local type family instances against all imported ones (authored by rwbarton).
Tue, Feb 14, 3:58 PM
bgamari updated subscribers of D3139: Try to fix boolformula into parsing "foo, bar, baz" into "And [foo, bar, baz]"instead of "And [foo, And [bar, baz]]".

Hi @ethercrow! Thanks for picking this up.

Tue, Feb 14, 3:04 PM
bgamari retitled D3139: Try to fix boolformula into parsing "foo, bar, baz" into "And [foo, bar, baz]"instead of "And [foo, And [bar, baz]]" from Try to fix boolformula into parsing "foo, bar, baz" into "And [foo, bar, baz]" instead of "And [foo, And [bar, baz]]" to Try to fix boolformula into parsing "foo, bar, baz" into "And [foo, bar, baz]"instead of "And [foo, And [bar, baz]]".
Tue, Feb 14, 2:57 PM
bgamari added inline comments to D3133: Improvements/bugfixes to signature reexport handling..
Tue, Feb 14, 12:15 PM
bgamari accepted D2992: Check local type family instances against all imported ones.

Lovely,

Tue, Feb 14, 12:12 PM
bgamari requested changes to D3114: Generate better fp abs for X86 and llvm..
Tue, Feb 14, 12:11 PM
bgamari added a comment to D3114: Generate better fp abs for X86 and llvm..

Bumping out of review queue for now.

Tue, Feb 14, 12:11 PM
bgamari abandoned D3138: Testing....

This has served its purpose.

Tue, Feb 14, 11:56 AM
bgamari accepted D3041: nucleic2: removed unused files & simplify Makefile.
Tue, Feb 14, 11:53 AM
bgamari requested changes to D3041: nucleic2: removed unused files & simplify Makefile.
Tue, Feb 14, 11:53 AM
bgamari accepted D3030: Fix nofib benchmarks on Windows..

Looks reasonable to me.

Tue, Feb 14, 11:52 AM
bgamari added inline comments to D810: Ensure that Literals are in range.
Tue, Feb 14, 11:35 AM
bgamari accepted D3125: Make deSugarExpr use runTcInteractive.

Looks fine to me.

Tue, Feb 14, 11:18 AM
bgamari added a comment to D3123: Subtyping for roles in signatures..

Don't we also need to document how roles are merged?

Tue, Feb 14, 11:16 AM