Re: MSYS2 package for GHC 7.10.1
Wow, this sounds great! Just to clarify - this would still be a mingw-w64 build and not require the msys2 DLLs, correct? Thanks, Yitz On May 21, 2015 23:53, David Macek david.mace...@gmail.com wrote: With the helpful pointers from ezyang on IRC, I pushed this a bit forward. I converted most of the patches into more reasonable commits including short descriptions and created a git branch for it. See https://github.com/ghc/ghc/compare/ghc-7.10.1-release...elieux:msys2-pkgbuild . As mentioned previously, the changes should be uncontroversial except for two big changes: removing bundled mingw, perl and touchy and changing the directory layout. While the directory layout change is mostly self-contained (barring any tools hardcoding ..\lib), the bundled dependency removal will required major changes to the build process. My proposals follow. For hacking on GHC == 1. Get MSYS2, update and install dependencies (including the bootstrapping ghc that would come as a MSYS2 package) 2. Get a GHC repository ready 3. Hack, hack, hack 4. Build and test as usual 5. GOTO 3 Alternatively, this could be replaced with a makepkg-based flow: 1. Get MSYS2, update and install dependencies (including the bootstrapping ghc that would come as a MSYS2 package) 2. Get a mingw-w64-ghc-git PKGBUILD 3. $ makepkg-mingw --nobuild # clone the repositories 4. Go to src/ghc and hack, hack, hack 5. $ makepkg-mingw --noextract --noprepare --noarchive # build and test 6. GOTO 4 For binary release == Phase 1: pacman package. This can be done in coordination with the MSYS2 maintainers, or a separate GHC-owned pacman repository can be created. 1. Get MSYS2, update and install dependencies (including the bootstrapping ghc that would come as a MSYS2 package) 2. Update the mingw-w64-ghc PKGBUILD to point to the new source release 3. $ makepkg-mingw # build a package 4. Upload the package to a pacman repository Phase 2: stand-alone bindist 1. Download the package and its dependencies 2. Extract them into a temporary directory 3. Create a tarball or an installer from that 4. Upload to GHC servers This is essentially what the new Git for Windows does (and what some other projects that use MSYS2 as their build environment do). -- David Macek ___ ghc-devs mailing list ghc-devs@haskell.org http://mail.haskell.org/cgi-bin/mailman/listinfo/ghc-devs ___ ghc-devs mailing list ghc-devs@haskell.org http://mail.haskell.org/cgi-bin/mailman/listinfo/ghc-devs
RE: [GHC] #10370: Compile time regression in OpenGLRaw
Austin: I have validated, but I am now going on holiday for a week. Can you monitor and revert if there's a major problem? Simon | -Original Message- | From: ghc-tickets [mailto:ghc-tickets-boun...@haskell.org] On Behalf | Of GHC | Sent: 22 May 2015 15:21 | Cc: ghc-tick...@haskell.org | Subject: Re: [GHC] #10370: Compile time regression in OpenGLRaw | | #10370: Compile time regression in OpenGLRaw | -+ | -- | -+--- | Reporter: michalt | Owner: | Type: bug | Status: new | Priority: normal| Milestone: | Component: Compiler | Version: | 7.10.1 |Resolution:|Keywords: | Operating System: Unknown/Multiple |Architecture: | Type of failure: Compile-time | Unknown/Multiple |performance bug| Test Case: |Blocked By:|Blocking: | Related Tickets:| Differential Revisions: | -+ | -- | -+--- | | Comment (by simonpj): | | The commit messages tell the story. Reid's example in comment:3 | showed up TWO separate, substantial non-linear performance holes in | GHC, both of which I have now fixed. Reid, your example was | amazingly helpful. | | '''Austin''' I have not added a `perf/compiler` test case. Could you | do so please? I wasn't sure that a 20-second compile time was | acceptable. | I'm leaving the ticket open for that reason. | | Simon | | -- | Ticket URL: http://ghc.haskell.org/trac/ghc/ticket/10370#comment:13 | GHC http://www.haskell.org/ghc/ | The Glasgow Haskell Compiler | ___ | ghc-tickets mailing list | ghc-tick...@haskell.org | http://mail.haskell.org/cgi-bin/mailman/listinfo/ghc-tickets ___ ghc-devs mailing list ghc-devs@haskell.org http://mail.haskell.org/cgi-bin/mailman/listinfo/ghc-devs
Re: [GHC] #10370: Compile time regression in OpenGLRaw
Will do. On Fri, May 22, 2015 at 9:23 AM, Simon Peyton Jones simo...@microsoft.com wrote: Austin: I have validated, but I am now going on holiday for a week. Can you monitor and revert if there's a major problem? Simon | -Original Message- | From: ghc-tickets [mailto:ghc-tickets-boun...@haskell.org] On Behalf | Of GHC | Sent: 22 May 2015 15:21 | Cc: ghc-tick...@haskell.org | Subject: Re: [GHC] #10370: Compile time regression in OpenGLRaw | | #10370: Compile time regression in OpenGLRaw | -+ | -- | -+--- | Reporter: michalt | Owner: | Type: bug | Status: new | Priority: normal| Milestone: | Component: Compiler | Version: | 7.10.1 |Resolution:|Keywords: | Operating System: Unknown/Multiple |Architecture: | Type of failure: Compile-time | Unknown/Multiple |performance bug| Test Case: |Blocked By:|Blocking: | Related Tickets:| Differential Revisions: | -+ | -- | -+--- | | Comment (by simonpj): | | The commit messages tell the story. Reid's example in comment:3 | showed up TWO separate, substantial non-linear performance holes in | GHC, both of which I have now fixed. Reid, your example was | amazingly helpful. | | '''Austin''' I have not added a `perf/compiler` test case. Could you | do so please? I wasn't sure that a 20-second compile time was | acceptable. | I'm leaving the ticket open for that reason. | | Simon | | -- | Ticket URL: http://ghc.haskell.org/trac/ghc/ticket/10370#comment:13 | GHC http://www.haskell.org/ghc/ | The Glasgow Haskell Compiler | ___ | ghc-tickets mailing list | ghc-tick...@haskell.org | http://mail.haskell.org/cgi-bin/mailman/listinfo/ghc-tickets ___ ghc-devs mailing list ghc-devs@haskell.org http://mail.haskell.org/cgi-bin/mailman/listinfo/ghc-devs -- Regards, Austin Seipp, Haskell Consultant Well-Typed LLP, http://www.well-typed.com/ ___ ghc-devs mailing list ghc-devs@haskell.org http://mail.haskell.org/cgi-bin/mailman/listinfo/ghc-devs
Re: MSYS2 package for GHC 7.10.1
On 22. 5. 2015 15:58, Yitzchak Gale wrote: Wow, this sounds great! Just to clarify - this would still be a mingw-w64 build and not require the msys2 DLLs, correct? Correct. -- David Macek smime.p7s Description: S/MIME Cryptographic Signature ___ ghc-devs mailing list ghc-devs@haskell.org http://mail.haskell.org/cgi-bin/mailman/listinfo/ghc-devs