- User Since
- Jan 8 2015, 11:07 PM (110 w, 5 d)
Sorry for responding so late on this. I'll also give my 👍 on this to provide one more vote of confidence from a CLC member.
Well that was fast... process-1.6 is now released, so this is technically out of date.
This has been superseded by http://git.haskell.org/ghc.git/commit/240b43e4969595418bb3c0d1519111be59572672, no?
Fri, Feb 10
- Accept improved error messages
Thu, Feb 9
- Address Richard's comment
Template Haskell pretty-printing is fixed, so this looks good to me.
Rebase on top of master
Thanks, @simonpj! We took a winding path to get here, but I think this finally does what it advertised. I've updated the comments and Note accordingly.
- It works!
Unfortunately, removing checkNoErrs doesn't change anything. I tried with and without the insolubleWC check. For posterity's sake, I've uploaded a Diff without checkNoErrs.
- Don't use checkNoErrs
Well I'm happy that we can delete a bunch of code, but sadly, doing so does not fix the testsuite failures (in fact, even more tests fail now).
- Simon's suggestions
Wed, Feb 8
Sorry, I meant arc patch D2961.
@simonpj, I feel I'm stuck again. I've tried two different things:
- in re simonpj
Tue, Feb 7
- Latest attempt
Mon, Feb 6
Looks great @dfeuer! One minor comment inline.
Sun, Feb 5
Assuming it validates, LGTM.
Thanks for working on this, @dfeuer!
Sat, Feb 4
Fri, Feb 3
Thu, Feb 2
@simonpj, I've updated mkMinimalBySCs to remove duplicate constraints, so now we don't need to call the simplifier twice.
- Don't call the simplifer twice needlessly
Oops, use right base commit
Make mkMinimalBySCs remove duplicate constraints
Wed, Feb 1
Ah! Then by all means, derive all the instances you find convenient.
I've no personal objections. Although I'm curious what the use case is—aren't these supposed to only ever appear at the type level?
Huh, I knew about the giant fixM block in the typechecker to tie an enormous recursive knot, but I never knew about forkM until now. Tricky stuff.
Tue, Jan 31
What is the status of this for GHC 8.2?
Mon, Jan 30
Rebase on top of master
Sun, Jan 29
Thanks @alanz, this is looking much better.
Sat, Jan 28
- Fix Haddocks
Fri, Jan 27
Also, a reminder to make a mention of this in the release notes, since I'll inevitably forget about all of these little differences in the linker between 8.0 and 8.2.
Urgh, there is a downside to this patch, though: this program (which legitimately requires -XTypeInType):
Thu, Jan 26
Actually, one tiny (but important) thing: can you mention this in the 8.2 release notes?
Oops, I didn't catch that you said you'd open a Trac issue in https://phabricator.haskell.org/D3031#88747. Splendid.
Indeed, you should at the very least:
Ah, I understand now. Given your description, I think this discrepancy might be acceptable. But we should advertise this in docs/users_guide/bugs.rst just in case (in addition to the release notes).
To be clear, does this require MSYS? Or does it work Cygwin with too?
I'm a bit lost. Can you give an example of a program which, using a version of GHC with this patch, would fail to compile but would work with GHCi?
Or maybe this should be punted till a proper implementation can be made for 8.4.1?
Any thoughts on this or would it not be useful?
I landed this patch (by hand) in http://git.haskell.org/nofib.git/commit/ce4b36b56727f1bdfc318fc0f7f4c4b192781537.
Wed, Jan 25
Tue, Jan 24
@bgamari, I fear you might have lost some commits during rebasing?
Looks great @dalaing, thanks. Can you also add a note to libraries/base/changelog.md about this change?
@erikd's the author, not me ;)
That fixes the issue for me! I now realize that I was using a build with DYNAMIC_GHC_PROGRAMS=NO, which was triggering the bug.
Mon, Jan 23
@simonpj, I've added a blurb at the end of Note [Default method type signatures must align] describing why the methods' contexts are allowed to differ.
- in re simonpj
Hopefully https://phabricator.haskell.org/D3002 resolves this issue.
Urk, it turns out that th-desugar has a test case that this commit broke:
Jan 23 2017
- mpickering's comments
Jan 22 2017
Good point @bgamari, I've added a Note (and references to it) in DsMeta.
- Add a Note to DsMeta