shlevy (Shea Levy)
User

Projects

User does not belong to any projects.

User Details

User Since
Nov 11 2016, 9:30 AM (88 w, 18 h)

Recent Activity

Feb 12 2018

shlevy updated the diff for D4388: Make cost centre symbol names deterministic..

Remove accidental shell.nix modification

Feb 12 2018, 8:58 PM
shlevy added a comment to D4388: Make cost centre symbol names deterministic..

@bgamari Fixed.

Feb 12 2018, 8:18 PM
shlevy updated the diff for D4388: Make cost centre symbol names deterministic..

Switch to FastStringEnv

Feb 12 2018, 8:17 PM
shlevy added a comment to D4388: Make cost centre symbol names deterministic..

I think we should update the relevant tickets (Trac Trac #4012 and Trac Trac #12935) with a description of the problem and the solution implemented in this patch.

Feb 12 2018, 1:11 PM

Feb 7 2018

shlevy added a comment to D4388: Make cost centre symbol names deterministic..

Sure. Given Test.hs:

Feb 7 2018, 7:25 AM

Feb 6 2018

shlevy created D4388: Make cost centre symbol names deterministic..
Feb 6 2018, 2:33 PM

Apr 30 2017

shlevy added a comment to D3496: Ignore ANN pragmas with no TH and no external interpreter..

Updated comments and warning. @RyanGlScott Still not clear to me how to add this test case, sorry, I can dig into this next weekend.

Apr 30 2017, 10:53 AM
shlevy updated the diff for D3496: Ignore ANN pragmas with no TH and no external interpreter..
  • Documentation and better warning
Apr 30 2017, 10:51 AM

Apr 25 2017

shlevy added a comment to D3496: Ignore ANN pragmas with no TH and no external interpreter..

@RyanGlScott I'm not familiar with the testsuite, can you point me to tests that specifically use stage1?

Apr 25 2017, 9:18 AM
shlevy created D3496: Ignore ANN pragmas with no TH and no external interpreter..
Apr 25 2017, 9:13 AM

Feb 28 2017

shlevy added a comment to D3232: iOS - Do *not* build terminfo..

Can we just do it with a flag, disabled by default?

Feb 28 2017, 7:11 AM
shlevy added a comment to D3232: iOS - Do *not* build terminfo..

What failure do you see?

Feb 28 2017, 7:00 AM
shlevy added a comment to D3232: iOS - Do *not* build terminfo..

Why are we disabling terminfo building on iOS? This works fine for me...

Feb 28 2017, 5:08 AM

Jan 14 2017

shlevy added a comment to D2960: Linker: iOS uses mach-o libraries..

No idea if it would pass app store approval, but it doesn't need to. The app that will use the linker (an iserv process) is only needed during development and thus can rely on the user having xcode and being able to test apps on their devices.

Jan 14 2017, 8:02 PM
shlevy added a comment to D2960: Linker: iOS uses mach-o libraries..

This is no longer true on recent versions.

Jan 14 2017, 4:39 PM

Jan 13 2017

shlevy updated the diff for D2970: Modify ForeignPtr documentation in light of plusForeignPtr.
  • Fix comment
Jan 13 2017, 6:11 AM

Jan 12 2017

shlevy added a comment to D2900: base: Add Foreign.ForeignPtr.plusForeignPtr..

https://phabricator.haskell.org/D2970

Jan 12 2017, 3:05 PM
shlevy retitled D2970: Modify ForeignPtr documentation in light of plusForeignPtr from to Modify ForeignPtr documentation in light of plusForeignPtr.
Jan 12 2017, 3:05 PM

Jan 11 2017

shlevy planned changes to D2960: Linker: iOS uses mach-o libraries..

This is not enough, more changes to linker needed.

Jan 11 2017, 8:23 PM
shlevy added a comment to D2960: Linker: iOS uses mach-o libraries..

Running the build now that will enable me to test

Jan 11 2017, 4:41 PM
shlevy retitled D2960: Linker: iOS uses mach-o libraries. from to Linker: iOS uses mach-o libraries..
Jan 11 2017, 2:19 PM
shlevy added a comment to D2900: base: Add Foreign.ForeignPtr.plusForeignPtr..

I'm not clear on what kind of documentation is being requested here? Happy to add whatever is needed.

Jan 11 2017, 9:28 AM
shlevy added a comment to D2900: base: Add Foreign.ForeignPtr.plusForeignPtr..

The kern.devstat.all FreeBSD sysctl fills the output buffer with a long (representing the generation number) followed by an array of struct devstat, and I want to expose that array using Data.Vector.Storable.unsafeFromForeignPtr. That function provides an offset argument, but that offset is in multiples of the underlying Storable type, not bytes.

Jan 11 2017, 9:28 AM

Jan 4 2017

shlevy abandoned D2902: Add atomicReadIntOffAddr# primop.
Jan 4 2017, 11:10 AM
shlevy added a comment to D2902: Add atomicReadIntOffAddr# primop.

I'm fine with a barrier solution as well, should I switch to that?

Jan 4 2017, 9:38 AM

Jan 3 2017

shlevy requested review of D2902: Add atomicReadIntOffAddr# primop.
Jan 3 2017, 7:00 AM

Jan 2 2017

shlevy added a comment to D1086: Accept next-docstrings on GADT constructors..

Ah, yes, I do, but I was thinking of it as function arguments due to the syntax. I'll follow that one.

Jan 2 2017, 8:14 PM
shlevy added a comment to D1086: Accept next-docstrings on GADT constructors..

What needs to be done to allow documentation of specific arguments of a particular constructor?

Jan 2 2017, 11:37 AM

Dec 29 2016

shlevy added a comment to D2902: Add atomicReadIntOffAddr# primop.

Not sure I got the casting right, but my dumb (single-threaded) tests worked right.

Dec 29 2016, 12:54 PM
shlevy updated the diff for D2902: Add atomicReadIntOffAddr# primop.
  • Add atomicReadInt{8,16,32,64}OffAddr# primops
Dec 29 2016, 12:50 PM

Dec 28 2016

shlevy added a comment to D2902: Add atomicReadIntOffAddr# primop.

I also want to add this for all of the signed integral types (my actual use case is CInt, but I wanted to start with this to be sure I understand how all of this ties together.

Dec 28 2016, 3:41 PM
shlevy retitled D2902: Add atomicReadIntOffAddr# primop from to Add atomicReadIntOffAddr# primop.
Dec 28 2016, 3:40 PM
shlevy added a comment to D2900: base: Add Foreign.ForeignPtr.plusForeignPtr..

test.hs:

{-# LANGUAGE MagicHash #-}
module Main where
Dec 28 2016, 10:21 AM
shlevy updated the diff for D2900: base: Add Foreign.ForeignPtr.plusForeignPtr..
  • base: Add some missing documentation for plusForeignPtr
Dec 28 2016, 9:07 AM

Dec 27 2016

shlevy retitled D2900: base: Add Foreign.ForeignPtr.plusForeignPtr. from to base: Add Foreign.ForeignPtr.plusForeignPtr..
Dec 27 2016, 4:05 PM

Dec 19 2016

shlevy requested review of D2884: Allow use of the external interpreter in stage1..

https://ghc.haskell.org/trac/ghc/ticket/13008

Dec 19 2016, 6:41 PM
shlevy retitled D2884: Allow use of the external interpreter in stage1. from GHCi: Fix build with GHC 7.10.x on Windows to Allow use of the external interpreter in stage1..
Dec 19 2016, 6:41 PM
shlevy added a comment to D2884: Allow use of the external interpreter in stage1..

https://github.com/shlevy/ghc/commit/726f4011540c0c3bb0ca6eed0e5f1233788c99ae is the interesting diff, the rest is just a revert of a revert.

Dec 19 2016, 4:23 PM
shlevy retitled D2884: Allow use of the external interpreter in stage1. from to GHCi: Fix build with GHC 7.10.x on Windows.
Dec 19 2016, 4:15 PM
shlevy added inline comments to rGHC4905b83a2d44: Remote GHCi, -fexternal-interpreter.
Dec 19 2016, 5:16 AM

Dec 16 2016

shlevy added a comment to D2826: Allow use of the external interpreter in stage1..

Unexpected test success is known and already fixed on master, unrelated to this.

Dec 16 2016, 6:45 PM

Dec 15 2016

shlevy updated the diff for D2826: Allow use of the external interpreter in stage1..
  • Fix comment
  • Rebase fixups
Dec 15 2016, 4:21 PM
shlevy added inline comments to rGHC4905b83a2d44: Remote GHCi, -fexternal-interpreter.
Dec 15 2016, 10:36 AM
shlevy updated the diff for D2826: Allow use of the external interpreter in stage1..
  • Minor formatting fixes
Dec 15 2016, 10:20 AM
shlevy added inline comments to D2826: Allow use of the external interpreter in stage1..
Dec 15 2016, 10:18 AM
shlevy added a comment to D2826: Allow use of the external interpreter in stage1..

@mpickering The unexpected test success on the OS X build is unrelated to this change and already fixed in master. Any other concerns here?

Dec 15 2016, 8:27 AM
shlevy added inline comments to rGHC4905b83a2d44: Remote GHCi, -fexternal-interpreter.
Dec 15 2016, 8:21 AM

Dec 14 2016

shlevy updated the diff for D2826: Allow use of the external interpreter in stage1..
  • Fix CPP macro for GHC.Stack.CCS
Dec 14 2016, 7:08 AM
shlevy added inline comments to D2826: Allow use of the external interpreter in stage1..
Dec 14 2016, 6:16 AM

Dec 13 2016

shlevy abandoned D2838: iserv: Communicate over stdio..

Ah, fair point, I was thinking only about the TH case, obviously this will break common GHCI uses. Damn!

Dec 13 2016, 2:08 PM
shlevy added a comment to D2838: iserv: Communicate over stdio..

OK, tests pass.

Dec 13 2016, 1:54 PM
shlevy updated the diff for D2838: iserv: Communicate over stdio..

Missed variable rename

Dec 13 2016, 1:40 PM
shlevy updated the diff for D2838: iserv: Communicate over stdio..

Add missing import

Dec 13 2016, 1:36 PM
mpickering awarded D2838: iserv: Communicate over stdio. a Like token.
Dec 13 2016, 1:03 PM
shlevy updated the diff for D2838: iserv: Communicate over stdio..

Fix bogus reference to command line fd numbers

Dec 13 2016, 12:58 PM
shlevy added a comment to D2838: iserv: Communicate over stdio..

Tests still running locally, just wanted to throw this up for discussion now.

Dec 13 2016, 12:50 PM
shlevy retitled D2838: iserv: Communicate over stdio. from to iserv: Communicate over stdio..
Dec 13 2016, 12:50 PM
shlevy updated the diff for D2826: Allow use of the external interpreter in stage1..
  • Enable building ghci with ghc 7.10.
Dec 13 2016, 11:07 AM
shlevy updated the diff for D2826: Allow use of the external interpreter in stage1..
  • Enable building ghci with ghc 7.10.
Dec 13 2016, 11:01 AM

Dec 12 2016

shlevy added a comment to D2826: Allow use of the external interpreter in stage1..

Failing now due to missing Generic instance for ExitCode. Can we just depend on >=8.0.1 in master?

Dec 12 2016, 6:08 PM
shlevy added a comment to D2826: Allow use of the external interpreter in stage1..

@simonmar I'm running into a bunch of "Could not find module ‘GHC.Stack.CCS’" building the ghci lib with 7.10.3 and , all of them so far are compiling if I change that to GHC.Stack. Are there important semantic differences here with previous GHC versions?

Dec 12 2016, 6:02 PM
shlevy added a comment to D2826: Allow use of the external interpreter in stage1..

Ah, the ghci library as a dependency on base 4.9.* and this adds ghci as a stage0 package. @simonmar is there any reason ghci couldn't be built with older base? Or can we bump our requirements to a more recent ghc?

Dec 12 2016, 8:23 AM
shlevy added inline comments to D2826: Allow use of the external interpreter in stage1..
Dec 12 2016, 7:57 AM
shlevy added a comment to D2826: Allow use of the external interpreter in stage1..

I haven't tried --interactive, and in fact I didn't change ghc-bin.cabal so I doubt it will work. I have tried template-haskell though!

Dec 12 2016, 7:56 AM

Dec 11 2016

shlevy added a comment to rGHCb7e88ee0d87f: Reduce the size of string literals in binaries..

https://phabricator.haskell.org/D2831

Dec 11 2016, 6:48 PM
shlevy retitled D2831: Fix pprCLabel on platforms without native codegen. from to Fix pprCLabel on platforms without native codegen..
Dec 11 2016, 6:47 PM
shlevy retitled D2830: Build terminfo on iOS. from to Build terminfo on iOS..
Dec 11 2016, 6:43 PM
shlevy added a comment to rGHC7036fde9df61: Overhaul of Compact Regions (#12455).

Never mind: https://phabricator.haskell.org/rGHCb7e88ee0d87f41cf1d8aba62aa44d5bf0a7404ad#56412

Dec 11 2016, 5:40 PM
shlevy added a comment to rGHCb7e88ee0d87f: Reduce the size of string literals in binaries..

This is because "cGhcWithNativeCodeGen" is false for iOS. The panic just seems bogus, right?

Dec 11 2016, 5:38 PM
shlevy raised a concern with rGHCb7e88ee0d87f: Reduce the size of string literals in binaries..

I hit the added panic here, building a cross-compiler targeting iOS:

Dec 11 2016, 5:34 PM
shlevy accepted rGHC52222f9bf705: Detect color support.

OK, terminfo seems to build fine on iOS now, I'm hitting another issue but once that goes fine I'll submit a patch for my concern.

Dec 11 2016, 5:29 PM
shlevy added a comment to rGHC7036fde9df61: Overhaul of Compact Regions (#12455).

Not sure if this is the culprit, but when trying to build an iOS cross compiler I'm getting:

Dec 11 2016, 5:22 PM
shlevy added a comment to rGHC52222f9bf705: Detect color support.

Hmm, https://ghc.haskell.org/trac/ghc/ticket/7700 suggests that this was added because iOS doesn't have termios, but it seems to now. Perhaps the enabling terminfo on ios is actually the right decision, I'll submit a differential if so

Dec 11 2016, 5:00 PM
shlevy raised a concern with rGHC52222f9bf705: Detect color support.

This breaks cross-compilation to iOS. We should either include terminfo in PACKAGES_STAGE1 for iOS (testing if this works now, but presumably it was disabled for a reason...) or turn off the terminfo flag in stage1 for iOS

Dec 11 2016, 4:56 PM
shlevy added a comment to rGHC1732d7ac43ca: Define thread primitives if they're supported..

Can you see what happens if you run with this reverted?

Dec 11 2016, 3:42 PM
shlevy retitled D2826: Allow use of the external interpreter in stage1. from to Allow use of the external interpreter in stage1..
Dec 11 2016, 2:28 PM
shlevy updated subscribers of rGHC1732d7ac43ca: Define thread primitives if they're supported..

@nomeata Are the benchmarks run against a non-threaded RTS? If it's threaded, it's almost certainly not this commit. If it's non-threaded, it's still probably not this commit, but note that we compile a number of unusued functions into the RTS in that case and it may be the linker is not as smart as @simonmar hoped and that's somehow bloating final executable size.

Dec 11 2016, 2:20 PM

Dec 10 2016

shlevy retitled D2823: Don't have CPP macros expanding to 'defined'. from to Don't have CPP macros expanding to 'defined'..
Dec 10 2016, 7:12 PM

Nov 28 2016

shlevy added inline comments to D2712: Define thread primitives if they're supported..
Nov 28 2016, 10:19 AM

Nov 26 2016

shlevy added a comment to D2712: Define thread primitives if they're supported..

Tests pass

Nov 26 2016, 7:50 AM
shlevy added a comment to D2712: Define thread primitives if they're supported..

Running tests

Nov 26 2016, 7:04 AM
shlevy retitled D2712: Define thread primitives if they're supported. from Define thread primitives if USE_PTHREAD_FOR_ITIMER to Define thread primitives if they're supported..
Nov 26 2016, 7:04 AM
shlevy updated the diff for D2712: Define thread primitives if they're supported..

Define thread primitives if they're supported.

Nov 26 2016, 7:03 AM

Nov 24 2016

shlevy added a comment to D2712: Define thread primitives if they're supported..

Tests pass

Nov 24 2016, 11:28 AM
shlevy added inline comments to D2712: Define thread primitives if they're supported..
Nov 24 2016, 11:12 AM
shlevy added a comment to D2712: Define thread primitives if they're supported..

Running local tests

Nov 24 2016, 9:41 AM
shlevy updated the diff for D2712: Define thread primitives if they're supported..

Define thread primitives if USE_PTHREAD_FOR_ITIMER

Nov 24 2016, 9:41 AM

Nov 21 2016

shlevy requested review of D2712: Define thread primitives if they're supported..
Nov 21 2016, 12:53 PM
shlevy added inline comments to D2712: Define thread primitives if they're supported..
Nov 21 2016, 12:52 PM

Nov 16 2016

shlevy requested review of D2712: Define thread primitives if they're supported..

@simonmar's requested change was already in the patch

Nov 16 2016, 10:42 AM
shlevy added inline comments to D2712: Define thread primitives if they're supported..
Nov 16 2016, 8:59 AM

Nov 15 2016

shlevy added a comment to D2712: Define thread primitives if they're supported..

Local tests succeed.

Nov 15 2016, 12:39 PM
shlevy added a comment to D2712: Define thread primitives if they're supported..

Tests still running locally.

Nov 15 2016, 12:07 PM
shlevy updated the diff for D2712: Define thread primitives if they're supported..

Define thread primitives if USE_PTHREAD_FOR_ITIMER

Nov 15 2016, 12:06 PM
shlevy updated the diff for D2713: Pass autoconf triplets to sub-project configures.

Pass autoconf triplets to sub-project configures

Nov 15 2016, 12:04 PM
shlevy updated the diff for D2710: Don't use mmap symbols when !RTS_LINKER_USE_MMAP.

Don't use mmap symbols when !RTS_LINKER_USE_MMAP

Nov 15 2016, 11:44 AM
shlevy added a comment to D2711: Storage.c: Pass a size to sys_icache_invalidate.

@bgamari Just happened to see the warning scroll by when looking at the logs for something else :)

Nov 15 2016, 11:38 AM
shlevy added a comment to D2710: Don't use mmap symbols when !RTS_LINKER_USE_MMAP.

Local tests pass

Nov 15 2016, 8:31 AM
shlevy retitled D2712: Define thread primitives if they're supported. from Define sync primitives if USE_PTHREAD_FOR_ITIMER to Define thread primitives if USE_PTHREAD_FOR_ITIMER.
Nov 15 2016, 7:42 AM
shlevy updated the diff for D2712: Define thread primitives if they're supported..

Define thread primitives if USE_PTHREAD_FOR_ITIMER

Nov 15 2016, 7:42 AM