
On 02/25/2015 10:57 AM, Neil Mitchell wrote:
When you say "change is on the 7.10 branch", you mean a fix for Haddock, or changes to Cabal?
To Haddock.
There are also other issues with the --hoogle output, the newest regression being that constructors are wrapped in square brackets, for example:
[Nothing] :: Maybe a
Generally, the --hoogle output from Haddock keeps going downhill. When I first implemented it, there was a test suite, but that seems to no longer be catching issues.
I'm aware and really want to fix these issues (Hoogle is great!) but as always what little manpower becomes available tends to go into showstopper bugs instead as apparent by the sad commit list in Haddock… I think I will try to jump into Hoogle problems at start of March and just ignore other stuff. As always, my apologies.
Thanks, Neil
On Wed, Feb 25, 2015 at 5:47 AM, Mateusz Kowalczyk
wrote: On 02/25/2015 05:41 AM, Michael Snoyman wrote:
I'm not really able to follow the details of this, but I wanted to raise to everyone's attention a serious bug with the current GHC 7.10 RC, Cabal 1.22, and/or Haddock. Currently, running `cabal haddock --hoogle` does not work. There seem to be two different issues open about it:
https://github.com/haskell/haddock/issues/361 https://github.com/haskell/cabal/issues/2297
I really don't understand the issues here, but I'd claim that the severity of this should probably be a blocker for a GHC 7.10 release.
I think only work needs to go into cabal now. GHCs point of concern here is that it ships with a patched cabal for this when the patch does get crafted rather than only dumped onto the master branch. It's already fixed on Haddock side and I believe the change is on 7.10 branch.
So I guess this is just a FYI that something is broken and we're awaiting cabal patch.
-- Mateusz K. _______________________________________________ ghc-devs mailing list ghc-devs@haskell.org http://mail.haskell.org/cgi-bin/mailman/listinfo/ghc-devs
-- Mateusz K.