Hacker News new | past | comments | ask | show | jobs | submit login

Nope. I also strongly recommend against using the normal "brew" formulas for haskell tools, its only by PROACTIVE efforts by mac using haskellers that the standard brew aren't a complete disaster (or at least, wont' be a disaster, soonish).

I strongly recommend using standard haskell binary distributions, such as those prepared by GHC HQ, or the Haskell Platform folks, or perhaps any brew Tap that Darin Morrison or Luke Ianni author. Do not use brew for haskell tools (at least until they actually follow Darin Morrison's word to the T, because they do janky stuff for haskell.)




Hmm. Well, looking at the Github repo activity, Darin is actively involved in the GHC formulae these days. In fact he's actively involved in getting Homebrew's GHC formula prepped for OS X Mavericks, which you note as an upcoming issue below. So that's encouraging.

And while the stability of GHC/Haskell Platform was indeed questionable during 2012.2.0.0 and the beginning of 2012.4.0.0, they cleaned it up as time went along and I haven't noticed any issues with it for the past 4 or 5 months. You make a comment below about Haskell Platform being associated with specific GHC releases, and AFAIK Homebrew actually freezes GHC at versions compatible with the current release of Haskell Platform now. The reason I prefer to use the package manager instead of the pre-distributed binaries is because it provides for a deterministic and (theoretically) testable/repeatable upgrade path.


1) you can very easily have multiple ghc's installed and have them not clobber each other 2) use the binary packages if you want a default that just works. If you look at the relevant ghc ticket on brew, they're very keen on modifying the formulae against the recommendations of Darin very very easily. 3) I use brew for many things, and even donated quite a bit to their kickstarter, but I do not trust brew to do Haskell right.


You are right.

I started with haskell-platform from brew and ended up with a broken ghc build.

"inplace/bin/ghc-stage1" -static -H32m -O -Iincludes -Iincludes/dist -Iincludes/dist-derivedconstants/header -Iincludes/dist-ghcconstants/header -Irts -Irts/dist/build -DCOMPILING_RTS -package-name rts -dcmm-lint -i -irts -irts/dist/build -irts/dist/build/autogen -Irts/dist/build -Irts/dist/build/autogen -O2 -c rts/Apply.cmm -o rts/dist/build/Apply.o /usr/local/bin/llc: : error: unable to get target for 'arm-apple-darwin10', see --version and --triple. make[1]: * [rts/dist/build/Apply.o] Error 1

I am starting over with the binary GHC distribution.


yeah, the brew folks don't understand the point of haskell platform, and honestly botch it. Their GHC formula should have something like ---add-Haskell-Platform-if-this-version-has-it (but a shorted name for it), because haskell platform is tied to specific GHC versions, rather than as another brew formula/package.

Haskell platform is tied to very very specific GHC releases, and have specially prepared and tested installers for each major OS target. Brew throws this all away and treats it like a simplistic "nice things" wrapper, when in fact its actually 1+ months of engineering time a year for Mark Lenctzer (and a handful of others) to make sure that everything JUST WORKS™ for new haskellers when they install GHC and tools on a machine.

also, for near term sanity, stick to Xcode 4.6 if you're using GHC 7.6

Without some hacky tricks i'm going to help the haskell platform folks work out, GHC 7.8 is the only GHC version that will play nice with OS X mavericks/Xcode 5. Hopefully with the tricks Luke Iannini and I have cooked up, GHC 7.6 in this fall's Haskell Platform release will work very nicely on Mavericks / Xcode 5. (but it will have to be a different haskell platform install than for systems pre Xcode 5)




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: