Add compatibility to GHC 7.8.x #45
Conversation
|
Great work, thank you. I'm a bit sleepy right now, so I will return to this tomorrow to check and merge. While you're at it, please amend that last commit and also edit “(< 7.10)” → “(< 7.8)” (that's in the section “Megaparsec vs Parsec”) changing description of the commit accordingly. Yep, we'll need to make Travis test GHC 7.8 too, but it's trivial. I'm not sure whether we should use custom Prelude and real benefits of such solution. I don't find conditional imports of |
This patch introduces compatibility to base-4.7.0.x. It was tested on Win 8.1 x86_64, using GHC 7.8.4. It mainly consists of a bunch of #if !MIN_VERSION(4,8,0) ... #endif additions and a lower bound on base in the cabal file as well as a general introduction of the CPP extension via default-extensions. It also removes a potential error source in tests/Util.hs, since the backslash in /=\ can lead to strange quirks on certain systems (backslash and newline only separated by whitespace). Other, squashed commits: - Remove 'recent version of base' from Readme - Change necessary version of GHC
Yeah, the conditional imports aren't that bad. However, the conditional imports/redefines on base-4.6.x.x... Well, see #46 :D.
Err, I think I've misread your comment as |
|
Yes, I meant only that last commit dealing with |
This PR basically introduces a bunch of CPP
#ifs to import additional operators fromControl.Applicative(in most cases) orData.Foldable(in one case). See #42 for the original issue.I didn't change the existing
import Control.Applicativelines, since they would lead to redundant import warnings on GHC 7.10.x, and instead put the#if MIN_VERSIONmakros at the end of all imports in all files. in all files that used only<$>in one place, I replaced<$>by infixfmapto keep the number of changes per file a little bit smaller.While I was at it, I also removed the "recent version of
base" comment in the readme, since it works now with an older one (by the way, base-4.6 compatibility seems possible, currently experimenting).Things left to do:
Remarks
/=\in tests/Utils.hs lead to some issues. I've added a comment, which will prevent backslash-before-newline trouble.Control.Applicativepart), but one should probably think of a less error prone strategy, like a custom prelude (so that there's only a single point of CPP).