- User Since
- Jun 6 2014, 6:07 AM (280 w, 4 d)
Mon, Sep 23
Aug 2 2019
Jul 26 2019
Apr 2 2019
Dec 30 2018
Moved to gitlab: https://gitlab.haskell.org/ghc/ghc/merge_requests/61
Moved to gitlab: https://gitlab.haskell.org/ghc/ghc/merge_requests/60
Dec 14 2018
Dec 10 2018
Dec 6 2018
The make buildsystem doesn't require me to re-run ./configure to use intree-gmp. I'm not sure on the build time impact rerunning ./configure has but maybe it should be taken into account?
Nov 27 2018
To me it sounds quite weird to "hide" the most important programs behind env vars (making them harder to discover and pass). I'd also like to note that I learned how to write this ac code by looking at how --with-ghc was implemented (which certainly is not optional :)).
Nov 8 2018
Nov 7 2018
Not sure if changes to FPTOOLS_ALEX in aclocal.m4 are necessary.
Oct 18 2018
@dfeuer any chance of a re-review?
Aug 26 2018
I think the sumarry has diverged from the code: "The trick is to use integer as the -this-unit-id" should be "use integer-wired-in".
Aug 16 2018
May 20 2018
- Revert "Mention where to find IOMode"
- Revert "Mention removal in changelog"
- Revert "Unexpose GHC.IO.IOMode in base"
May 15 2018
Related to the "community work", has someone put all code on hackage in a searchable database somewhere?
Pr to process: https://github.com/haskell/process/pull/129
- Mention where to find IOMode
Looks like process imports GHC.IO.IOMode, what's the process here? PR to process to import it from System.IO?
- Unexpose GHC.IO.IOMode in base
- Mention removal in changelog
@dfeuer can you clarify: I will leave GHC.IntWord64 (ghc-prim) as is currently (ie not-home) and remove GHC.IO.IOMode from exposed-modules?
@dfeuer, GHC.IntWord64 is from ghc-prim, that one should certainly not be removed from exposed-modules? (base imports it)
May 14 2018
- Remove OPTIONS_HADDOCK hide in favour for not-home (compiler)
(Gave the same treatment to ghc-prim)
- Remove HADDOCK_OPTIONS hide in favour for not-home
My rationale for this change is that hide limits exploration of base and therefore learning. Using not-home instead of hide we will generate documentation and there will be clickable links on the package homepage on hackage, while letting the home of ie CallStack be GHC.Stack and not GHC.Stack.Types.
Feb 1 2018
Jul 18 2017
- Fix Haddock notation
Jul 17 2017
- Correct spelling
Apr 25 2017
- Add Semigroup instance, move haddocks around
Apr 4 2017
I saw this suggestion in the original thread: https://mail.haskell.org/pipermail/libraries/2015-July/025978.html, perhaps something to consider?
Jan 22 2016
Dec 14 2015
- Remove linebreak
- Make sure that a lazy bind does what it should
I might want to add to the testcase that lazy binds work as they should.
- Add testcase
Thanks @thomie, that seems like it will do the trick!
Not sure how to test this. Is core stable enough to inspect in tests? But that doesn't seem particularly easy to do anyway...
Dec 8 2015
Nov 26 2015
Nov 17 2015
Nov 13 2015
- Strict: refer to note
Nov 12 2015
- Update expected GHC extensions
- make warnings on non-exhaustive pattern matches in let work
Nov 6 2015
Oct 29 2015
Todo: fix new error
- rebase again
- fix rebase
- strict: fix compilation
- strict: beautify users guide entry
- Strict: Use IdEnv to find corresponding global vars
- strict: fix patterns with no binders
- strict: add some examples to comment
- strict: share some code
- strict: add comments
- strict: expand note
Oct 16 2015
Mark a bunch of comments as done.
- Strict: update comments here and there
Oct 6 2015
- Strict: fix T4437
- Update users guide entry to rst
Sep 27 2015
- Strict: OrdList -> 
- Strict: add regression test for let-like binders
Sep 23 2015
You're welcome. It was nice to be able to help!
It seems the testsuite was fixed in rGHC79f57325dca4d1ce4601d01c4fab50f7bcfc9b9b.
Sep 22 2015
Also this line still appears
It seems output was changed by https://github.com/haskell/cabal/commit/6661945754d62fc6ad5e9fd68d17f472adfb44d0.
Both failing test-cases (https://phabricator.haskell.org/harbormaster/build/6301/) seem to be related to Cabal in some way...
This patch breaks the build:
- Strict: Accept new error output T6078
- Add Strict to expected GHC language extensions test
- add Strict to users guide
- add literal tags to docs
- strict toEnum error currently
- Strict: Desugar strict bindings working
- Strict: fix T9140
- add back opt_strict
- Working desugaring?
Sep 21 2015
Aug 21 2015
An observation, allowing polymorphic bang patterns might break some code: