- ghc-devs - Haskell.org

Parallel library
by Simon Peyton Jones 30 Jul '15

30 Jul '15

Simon Are you the maintainer of the 'parallel' library? The core libraries wiki page says yes: https://wiki.haskell.org/Library_submissions#The_Core_Libraries But parallel.cabal says "Maintainer: libraries(a)haskell.org<mailto:libraries@haskell.org>", which I thought we'd gotten rid of entirely. I ask because some should fix this: libraries/parallel/Control/Parallel/Strategies.hs:513:2: warning: Rule "parList/rseq" may never fire because 'rseq' might inline first Probable fix: add an INLINE[n] or NOINLINE[n] pragma for 'rseq' We need a pragma on 'rseq'. Simon

5 10

RE: [Diffusion] [Committed] rGHC24afe6d3aae1: Fix missing files
by Simon Peyton Jones 30 Jul '15

30 Jul '15

Sorry for all the noise recently from me. I _think_ we should be back to clean validation now Simon | -----Original Message----- | From: noreply(a)phabricator.haskell.org | [mailto:noreply@phabricator.haskell.org] | Sent: 30 July 2015 15:53 | To: Simon Peyton Jones | Subject: [Diffusion] [Committed] rGHC24afe6d3aae1: Fix missing files | | simonpj committed rGHC24afe6d3aae1: Fix missing files (authored by | simonpj). | | Fix missing files | | These two testsuite files were correct in my build tree but not in my | source tree; apologies. | | | BRANCHES | master | | AFFECTED FILES | /testsuite/tests/simplCore/should_compile/T7287.stderr | /testsuite/tests/stranal/should_compile/T10694.stdout | | USERS | simonpj (Author) | GHC - Testsuite (Auditor) | | COMMIT | https://phabricator.haskell.org/rGHC24afe6d3aae1 | | EMAIL PREFERENCES | https://phabricator.haskell.org/settings/panel/emailpreferences/ | | To: simonpj, GHC - Testsuite | Cc: thomie

1 0

ANNOUNCE: GHC version 7.10.2
by Ben Gamari 30 Jul '15

30 Jul '15

===================================================================== The (Interactive) Glasgow Haskell Compiler -- version 7.10.2 ===================================================================== The GHC Team is pleased to announce a new bug-fix release of GHC, 7.10.2. There have been a number of significant bug fixes since the 7.10.1 (with over 70 defects fixed). These include fixes affecting type-checker correctness, runtime stability, and compiler performance. For this reason we highly recommend that users of 7.10.1 upgrade quickly. Note that due to changes made in this release to GHC's optimizer, previously fragile-but-working rewrite rules may fail to fire. One example of this was found in the widely-used `text` package late in the release cycle, which manifested in long compilation times and poor code generation for `Text` literals (see [Trac #10528]). Users are advised to use text >=1.2.1.2 which includes more robust rewrite rules. The full release notes including a complete listing of the changes in this release can be found here, https://downloads.haskell.org/~ghc/7.10.2/docs/html/users_guide/release-7-1… [Trac #10528]: https://ghc.haskell.org/trac/ghc/ticket/10528 How to get it ~~~~~~~~~~~~~ Both binary and source tarballs of GHC itself are available on the release download page, https://www.haskell.org/ghc/download_ghc_7_10_2 For a more smoother, better integrated experience users are encouraged to find a Haskell distribution. While these alternatives may not yet include 7.10.2, they offer tools and libraries to quickly get users up and running, as well as potentially better integration with the host operating system and package manager. See, https://www.haskell.org/downloads for more details. Background ~~~~~~~~~~ Haskell is a standard lazy functional programming language. GHC is a state-of-the-art programming suite for Haskell. Included is an optimising compiler generating good code for a variety of platforms, together with an interactive system for convenient, quick development. The distribution includes space and time profiling facilities, a large collection of libraries, and support for various language extensions, including concurrency, exceptions, and foreign language interfaces (C, whatever). GHC is distributed under a BSD-style open source license. A wide variety of Haskell related resources (tutorials, libraries, specifications, documentation, compilers, interpreters, references, contact information, links to research groups) are available from the Haskell home page (see below), https://www.haskell.org/ Supported Platforms ~~~~~~~~~~~~~~~~~~~ The list of platforms supported by GHC and the people responsible for them can be found here, https://ghc.haskell.org/trac/ghc/wiki/Platforms https://ghc.haskell.org/trac/ghc/wiki/CodeOwners Ports to other platforms are possible with varying degrees of difficulty. The Building Guide describes how to go about porting to a new platform: https://ghc.haskell.org/trac/ghc/wiki/Building Developers ~~~~~~~~~~ If you enjoy using GHC, you will likely also enjoy contributing to it! We are always looking for new contributors. Instructions on accessing our source code repository, and getting started with hacking on GHC, are available from the GHC developer site, https://ghc.haskell.org/trac/ghc/ Mailing lists ~~~~~~~~~~~~~ We run mailing lists for GHC users and bug reports; to subscribe, use the web interfaces at https://www.haskell.org/mailman/listinfo/glasgow-haskell-users https://www.haskell.org/mailman/listinfo/glasgow-haskell-bugs There are several other haskell and ghc-related mailing lists on www.haskell.org; for the full list, see https://www.haskell.org/mailman/listinfo/ Some GHC developers hang out on #haskell on IRC, too: https://www.haskell.org/haskellwiki/IRC_channel Please report bugs using our bug tracking system. Instructions on reporting bugs can be found here: https://www.haskell.org/ghc/reportabug Hashes & Signatures ~~~~~~~~~~~~~~~~~ In https://downloads.haskell.org/~ghc/7.10.2/ you will find a signed copy of the SHA256 hashes for the tarballs, using my GPG key, Benjamin Gamari <ben(a)well-typed.com> Fingerprint: FFEB 7CE8 1E16 A36B 3E2D ED6F 2DE0 4D4E 97DB 64AD

3 6

Question about inliner behaviour with a small function
by Michael Walker 29 Jul '15

29 Jul '15

Hello, I managed to shrink a bizarre memory issue down to this (probably minimal) example: module Main where f :: [Int] -> Int f xs = length is where is = [ i | (_, i) <- pairs ys ] :: [(Int,Int)] ys = zip [0..] xs :: [(Int, Int)] {-# INLINE pairs #-} pairs xs = zip xs $ tail xs main :: IO () main = print $ f xs where xs = replicate 1000000 0 With the INLINE pragma, this allocates 264,049,584 bytes (compiled with `ghc -O2 -prof -fprof-auto inline.hs), without the pragma it allocates 336,049,512 bytes. Dropping the `main` definition (and renaming the module to "Foo") and examining the core, the key difference seems to be in how the list comprehension is compiled. With the INLINE pragma: Rec { Foo.f_go [Occ=LoopBreaker] :: [(Int, Int)] -> [(Int, Int)] -> [(Int, Int)] [GblId, Arity=2, Caf=NoCafRefs, Str=DmdType <S,1*U><L,1*U>] Foo.f_go = \ (ds_a14a :: [(Int, Int)]) (_ys_a14b :: [(Int, Int)]) -> case ds_a14a of _ [Occ=Dead] { [] -> GHC.Types.[] @ (Int, Int); : ipv_a14g ipv1_a14h -> case _ys_a14b of _ [Occ=Dead] { [] -> GHC.Types.[] @ (Int, Int); : ipv2_a14n ipv3_a14o -> GHC.Types.: @ (Int, Int) ipv2_a14n (Foo.f_go ipv1_a14h ipv3_a14o) } } end Rec } And without: Rec { Foo.f_go [Occ=LoopBreaker] :: [((Int, Int), (Int, Int))] -> [(Int, Int)] [GblId, Arity=1, Caf=NoCafRefs, Str=DmdType <S,1*U>] Foo.f_go = \ (ds_a13o :: [((Int, Int), (Int, Int))]) -> case ds_a13o of _ [Occ=Dead] { [] -> GHC.Types.[] @ (Int, Int); : y_a13t ys_a13u -> case y_a13t of _ [Occ=Dead] { (ds1_d12V, i_an2) -> GHC.Types.: @ (Int, Int) i_an2 (Foo.f_go ys_a13u) } } end Rec } Full core is available at https://gist.github.com/barrucadu/a59df62cd16074559e35 In the no-INLINE case, the list comprehension is compiled much like I would expect, it's walking down a zipped list of pairs and producing the result. The INLINE case is rather different, it looks like the `zip` has been inlined and deforestation has happened. This explains the difference in memory usage, a whole intermediary list has been skipped! I assume that GHC's analysis is determining `pairs` is too expensive to inline early enough to allow the further optimisation without the pragma, but the pragma forces it to happen earlier by marking it as really cheap. But `pairs` is a really small definition, syntactically. Why does the analysis consider it expensive? It is simply because it uses its argument multiple times in its body? And if it's not some sort of cost analysis, what's really going on? Thank for your time. -- Michael Walker (http://www.barrucadu.co.uk)

2 1

RE: warnings
by Simon Peyton Jones 28 Jul '15

28 Jul '15

Hopefully done now. Sorry about this -- really busy day S From: Simon Peyton Jones Sent: 28 July 2015 11:01 To: 'ghc-devs' Subject: RE: warnings Validating now, but have to go to London for a few hrs. Meanwhile use -Wwarn! From: Simon Peyton Jones Sent: 27 July 2015 17:08 To: ghc-devs Subject: warnings Yurgh. I committed a patch but my validate had -Wwarn by mistake. So there are warnings in base (and binary actually). I will sort them out tomorrow. Could revert and re-apply but doesn't seem worth it. Sorry Simon

2 2

Ideas for research
by Nikita Kartashov 28 Jul '15

28 Jul '15

Hello ghc-devs, I’m starting a PhD program in September and in need of a topic. Could you point me to some projects that still need to be implemented or even general areas which need to be explored? https://ghc.haskell.org/trac/ghc/wiki/ProjectSuggestions page is old and probably not up-to-date. I’m not shy of coding or math, and will be able to spend time reading papers to familiarise myself with the arising problems. With regards, Nikita Kartashov

2 1

RE: warnings
by Simon Peyton Jones 28 Jul '15

28 Jul '15

Validating now, but have to go to London for a few hrs. Meanwhile use -Wwarn! From: Simon Peyton Jones Sent: 27 July 2015 17:08 To: ghc-devs Subject: warnings Yurgh. I committed a patch but my validate had -Wwarn by mistake. So there are warnings in base (and binary actually). I will sort them out tomorrow. Could revert and re-apply but doesn't seem worth it. Sorry Simon

1 0

Trac is slow
by Simon Peyton Jones 28 Jul '15

28 Jul '15

Trac is terribly slow today. More than 60 seconds to load a single page. Help! Simon

1 0

RE: Deleting sync-all
by Tamar Christina 28 Jul '15

28 Jul '15

Hi Thomas, Sure I can take care of that part. I'll make a separate issue for it. I also think we should do the first part as well eventually (using mingw-w for for both x64 and x86_64) as that solves #10435 . Regards, Tamar ------------------------------ From: Thomas Miedema <thomasmiedema(a)gmail.com> Sent: ‎7/‎27/‎2015 18:25 To: Tamar Christina <lonetiger(a)gmail.com> Cc: Páli Gábor János <pali.gabor(a)gmail.com>; ghc-devs(a)haskell.org Subject: Re: Deleting sync-all Hi Tamar, feel free to take over that patch <https://phabricator.haskell.org/D339>, or extract the parts you like. It currently does (too) many things, which is part of the reason why it stalled. See my comments in #9218 <https://ghc.haskell.org/trac/ghc/ticket/9218>. Do not wait to get rid of the perl dependency first, because it is yet unclear how to proceed (#9832 <https://ghc.haskell.org/trac/ghc/ticket/9832>). Thomas On Mon, Jul 27, 2015 at 2:34 PM, Tamar Christina <lonetiger(a)gmail.com> wrote: > Hi > > There is a sorta in limbo differential from gintas that would have solved > this by doing it in the configure script. > > I liked that solution. As far as I know the patch was delayed because the > reviewers suggested we get rid of the perl dependency first. > > Regards, > Tamar > ------------------------------ > From: Thomas Miedema <thomasmiedema(a)gmail.com> > Sent: ‎7/‎27/‎2015 10:41 > To: Páli Gábor János <pali.gabor(a)gmail.com> > Cc: ghc-devs(a)haskell.org > Subject: Re: Deleting sync-all > > Hi Gábor, > > instead of changing `ghc.mk`, please change the builder script to also > run `git clone git://git.haskell.org/ghc-tarballs.git`. > > (Edit: while typing this, I saw you just did so) > > The benefits are: > * the windows specific part of creating a source distribution also gets > tested > * the release manager can run `make sdist` on any platform, and they'll > get a complete source distribution they can upload > > Thomas > > On Sun, Jul 26, 2015 at 8:36 PM, Páli Gábor János <pali.gabor(a)gmail.com> > wrote: > >> Hi Thomas, >> >> 2015-07-21 12:45 GMT+02:00 Thomas Miedema <thomasmiedema(a)gmail.com>: >> > To get the sources: >> > >> > git clone --recursive git://git.haskell.org/ghc.git >> > cd ghc >> > git clone git://git.haskell.org/ghc-tarballs.git # Windows only >> >> Apparently, that last line is also needed even on non-Windows systems >> when one wants to make the "sdist" target, at least as I have observed >> on my builders, e.g. [1]. >> >> So I have put together a small patch that aims to fix this -- mind >> you, I have not tested it on Windows, but theoretically it shall just >> work: >> >> --- a/ghc.mk >> +++ b/ghc.mk >> @@ -1183,9 +1183,14 @@ sdist-windows-tarballs: sdist-windows-tarballs-prep >> sdist-testsuite: sdist-testsuite-prep >> cd $(SRC_DIST_TESTSUITE_ROOT) && "$(TAR_CMD)" chf - >> $(SRC_DIST_BASE_NAME) 2> testsuite_log | $(TAR_COMP_CMD) >> -c > $(TOP)/$(SRC_DIST_TESTSUITE_TARBALL) >> >> +SDIST_TARGETS= sdist-ghc sdist-testsuite >> + >> +ifeq "$(Windows_Host)" "YES" >> +SDIST_TARGETS+= sdist-windows-tarballs >> +endif >> >> .PHONY: sdist >> -sdist : sdist-ghc sdist-windows-tarballs sdist-testsuite >> +sdist : $(SDIST_TARGETS) >> >> sdist-manifest : $(SRC_DIST_GHC_TARBALL) >> tar tjf $(SRC_DIST_GHC_TARBALL) | sed >> "s|^ghc-$(ProjectVersion)/||" | sort >sdist-manifest >> diff --git a/utils/haddock b/utils/haddock >> >> Are there any objections to commit this? >> >> Thanks, >> Gábor >> >> [1] http://haskell.inf.elte.hu/builders/freebsd-amd64-head/700/15.html >> > >

1 0

RE: Deleting sync-all
by Tamar Christina 27 Jul '15

27 Jul '15

Hi There is a sorta in limbo differential from gintas that would have solved this by doing it in the configure script. I liked that solution. As far as I know the patch was delayed because the reviewers suggested we get rid of the perl dependency first. Regards, Tamar ------------------------------ From: Thomas Miedema <thomasmiedema(a)gmail.com> Sent: ‎7/‎27/‎2015 10:41 To: Páli Gábor János <pali.gabor(a)gmail.com> Cc: ghc-devs(a)haskell.org Subject: Re: Deleting sync-all Hi Gábor, instead of changing `ghc.mk`, please change the builder script to also run `git clone git://git.haskell.org/ghc-tarballs.git`. (Edit: while typing this, I saw you just did so) The benefits are: * the windows specific part of creating a source distribution also gets tested * the release manager can run `make sdist` on any platform, and they'll get a complete source distribution they can upload Thomas On Sun, Jul 26, 2015 at 8:36 PM, Páli Gábor János <pali.gabor(a)gmail.com> wrote: > Hi Thomas, > > 2015-07-21 12:45 GMT+02:00 Thomas Miedema <thomasmiedema(a)gmail.com>: > > To get the sources: > > > > git clone --recursive git://git.haskell.org/ghc.git > > cd ghc > > git clone git://git.haskell.org/ghc-tarballs.git # Windows only > > Apparently, that last line is also needed even on non-Windows systems > when one wants to make the "sdist" target, at least as I have observed > on my builders, e.g. [1]. > > So I have put together a small patch that aims to fix this -- mind > you, I have not tested it on Windows, but theoretically it shall just > work: > > --- a/ghc.mk > +++ b/ghc.mk > @@ -1183,9 +1183,14 @@ sdist-windows-tarballs: sdist-windows-tarballs-prep > sdist-testsuite: sdist-testsuite-prep > cd $(SRC_DIST_TESTSUITE_ROOT) && "$(TAR_CMD)" chf - > $(SRC_DIST_BASE_NAME) 2> testsuite_log | $(TAR_COMP_CMD) > -c > $(TOP)/$(SRC_DIST_TESTSUITE_TARBALL) > > +SDIST_TARGETS= sdist-ghc sdist-testsuite > + > +ifeq "$(Windows_Host)" "YES" > +SDIST_TARGETS+= sdist-windows-tarballs > +endif > > .PHONY: sdist > -sdist : sdist-ghc sdist-windows-tarballs sdist-testsuite > +sdist : $(SDIST_TARGETS) > > sdist-manifest : $(SRC_DIST_GHC_TARBALL) > tar tjf $(SRC_DIST_GHC_TARBALL) | sed > "s|^ghc-$(ProjectVersion)/||" | sort >sdist-manifest > diff --git a/utils/haddock b/utils/haddock > > Are there any objections to commit this? > > Thanks, > Gábor > > [1] http://haskell.inf.elte.hu/builders/freebsd-amd64-head/700/15.html >

2 1