March 2013 - iPhone - Haskell.org

Here's exactly how to build the master-merge-ios branch
by Stephen Blackheath [to GHC-iPhone] 30 Aug '13

30 Aug '13

Stefan, Here is the exact procedure I just did to successfully build the GHC-iOS cross compiler: git clone git@github.com:ghc-ios/ghc.git ghc-ios cd ghc-ios git checkout master-merge-ios cp arm-apple-darwin10-bin/* ~/bin/ (<-- ~/bin is in my PATH) I edited ~/bin/arm-apple-darwin10-gcc to change the -march option to armv7, like this, because that's what my iPod touch has: TARGET_CFLAGS="-isysroot $TARGET_PLATFORM -march=armv7 -mcpu=cortex-a8 -mfpu=neon" git checkout master git remote set-url origin http://darcs.haskell.org/ghc.git/ ./sync-all --no-dph get wget http://hackage.haskell.org/trac/ghc/raw-attachment/ticket/7759/ios-patch-14… wget http://hackage.haskell.org/trac/ghc/raw-attachment/ticket/7760/ios-patch-15… cd libraries/base cat ../../ios-patch-14a-libraries-base.patch | patch -p1 cd ../../utils/hsc2hs/ cat ../../ios-patch-15a-against-hsc2hs-remove-HSC2HS_EXTRA-from-stage0-wrapper.patch | patch -p1 cd ../.. git remote set-url origin git@github.com:ghc-ios/ghc.git git checkout master-merge-ios cat > mk/build.mk HADDOCK_DOCS = NO BUILD_DOCBOOK_HTML = NO BUILD_DOCBOOK_PS = NO BUILD_DOCBOOK_PDF = NO SPLIT_OBJS = NO INTEGER_LIBRARY = integer-simple Stage1Only = YES DYNAMIC_GHC_PROGRAMS = NO GhcLibWays = v <ctrl-D> perl boot ./configure --target=arm-apple-darwin10 --prefix=/usr/local/ghc-ios-steve make When you get this: ghc-stage1: panic! (the 'impossible' happened) (GHC version 7.7.20130328 for arm-apple-ios): expectJust initTcInteractive just type 'make' again I am using: ghc-7.4.2 as my bootstrap ghc compiler thopminkringscrape:ghc-ios stephenblackheath$ llc --version Low Level Virtual Machine (http://llvm.org/) llvm version 3.0 Optimized build. Built Aug 10 2012 (18:51:37). Host: x86_64-apple-darwin12.3.0 Host CPU: i686 This is the head for master-merge-ios branch: commit 942b130ada7be3f608286da934026e2c162d9527 Merge: 97274b1 81d55a9 Author: Stephen Blackheath <..> Date: Thu Mar 28 23:51:02 2013 +1300 Merge branch 'master' into master-merge-ios > would you mind uploading somewhere a source tree that builds for you? Sorry - I'm not really able to do that at the moment. > you mean the architecture compiled for? armv7 vs armv7s? That's right. I have not actually run this version yet. Steve On 28/03/13 23:21, Stefan Kersten wrote: > hi steve, > > On 26 Mar 2013, at 20:36, "Stephen Blackheath [to GHC-iPhone]" <likeliest.complexions.stephen(a)blacksapphire.com> wrote: >> The rev you mentioned _definitely_ works. >> >> Check that you're using llvm version 3.0 or >= 3.2. Llvm version 3.1 is broken. > > i'm using llvm version 3.2, 3.0 doesn't compile anymore with the newest toolchain. still no joy, started from scratch and now having trouble getting as far as i already got. if you have the time and resources, would you mind uploading somewhere a source tree that builds for you? which osx+xcode+llvm are you using? > >> If it's not that, then it could be something to do with ARM variants. > > you mean the architecture compiled for? armv7 vs armv7s? > > many thanks + best, > sk >

8 11

iOS cross compile instructions now on GHC wiki / progress update
by Stephen Blackheath [to GHC-iPhone] 13 Aug '13

13 Aug '13

All, Cross-compiling GHC to iOS: http://hackage.haskell.org/trac/ghc/wiki/Building/CrossCompiling/iOS Progress update: I've now submitted all the patches as tickets to GHC. I've hit a regression with http://hackage.haskell.org/trac/ghc/ticket/7722 (this was working in the old 'ios' branch at http://github.com/ghc-ios) but this won't stop trivial programs running. Steve

7 30

Re: [Haskell-iPhone] Has anyone submitted an app using GHC-IOS?
by Robert Lorentz 02 May '13

02 May '13

Hi Jan thanks for the email. I'm replying to the list because it was a good question Of course the actual warnings wouldn't end up in a binary. However, the warnings are due to using linker options regarding disabling relocate-able code I believe (if I recall correctly - it's been a few weeks..). The resulting code works on devices - that is a fact. However, it is strongly advised to NOT use that linker option. This is evidenced by Apple docs I have seen (sorry no references handy) - and the fact that the compiler/linker complains when you use that linker option. This is enough of "gray area" that I am not 100% convinced that an app would be accepted. I understand that there is static analysis done automatically on apps being reviewed. (No reference). This analysis could potentially find that the linked binary had been created in such a manner and reject it What you say about the Haskell vs C source is true EXCEPT that the Haskell runtime must be compiled with these specific options. Generally speaking - it's a bug that the runtime can't just be compiled "normally". It can be fixed. But as far as I have heard from luke and Stephen they aren't working on it. So there's risk there by choosing to use this GHC iOS Things may be different than they were when I tried GHC iOS (December or November I think). I'm sure there are errors in some things I'm writing here, GHC iOS is a very casual thing for me, I'm barely involved and by no means am I trying to come across as someone that knows what's going on! I am very interested in seeing it succeed though I'm 90% sure the App Store would accept with no problems. But it's easy enough to try to be more certain. I will submit an app this week using GHC iOS and will report back to the group On Mar 2, 2013, at 7:29 PM, Jan Greve <0x4034e61(a)googlemail.com> wrote: > > Hi Robert, > > Why exactly would linking warnings pop up in an app store review? Those are not traceable in a binary. Also, it seems like the App Store approval is based on "look and feel", rather then good coding style. Even Apps with memory leaks are permitted sometimes - and that is something one can detect easily by using instruments alongside the functional testing. > > Plus, I don't see how a Haskell-sourced arm binary would differ from a C-based one. Thus, theoretically there should be no problem getting an App approved for the App Store that uses a Library written in Haskell. > > Cheers, Jan Greve > > Am 27.02.2013 um 17:41 schrieb Robert Lorentz <robert.lorentz(a)me.com>: > >> Curious if this would be approved in the App Store especially with respect to the warnings you get when linking. Has anyone succeeded? >> >> I would be willing to do this exercise with one of my apps (simply put in a rev that adds the hello world type example) - but if someone else has done it I won't waste my time >> _______________________________________________ >> iPhone mailing list >> iPhone(a)haskell.org >> http://www.haskell.org/mailman/listinfo/iphone >

2 2

build troubles with ghc-ios
by Stefan Kersten 26 Mar '13

26 Mar '13

hi all, i'm having a hard time building ghc from the ghc-ios ios branch. i am using the script that has been posted recently on this list for checking out the correct versions of library packages [1]. below is the error i get for rts/PrimOps.cmm (libraries/ghc-prim/GHC/Generics.hs fails with a similar error); this is on osx 10.8.3 with xcode 4.6.1. i've tried compiling with ghc 7.4.1 and 7.4.2, both built for i386 and x86_64, respectively. does this look familiar to anybody? thanks, sk [1] https://github.com/pjones/ghc-ios-util/blob/master/ghc/checkout.sh "inplace/bin/ghc-stage1" -H64m -O2 -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 -optc-O2 -c rts/PrimOps.cmm -o rts/dist/build/PrimOps.o /var/folders/qh/jdhbtx3x10jfrs0yq3qrfdx00000gn/T/ghc40705_0/ghc40705_0.s:3966:0: bad immediate value for offset (6012) /var/folders/qh/jdhbtx3x10jfrs0yq3qrfdx00000gn/T/ghc40705_0/ghc40705_0.s:3960:0: bad immediate value for offset (5992) /var/folders/qh/jdhbtx3x10jfrs0yq3qrfdx00000gn/T/ghc40705_0/ghc40705_0.s:3926:0: bad immediate value for offset (5920) /var/folders/qh/jdhbtx3x10jfrs0yq3qrfdx00000gn/T/ghc40705_0/ghc40705_0.s:3920:0: bad immediate value for offset (5900) /var/folders/qh/jdhbtx3x10jfrs0yq3qrfdx00000gn/T/ghc40705_0/ghc40705_0.s:3899:0: bad immediate value for offset (5828) /var/folders/qh/jdhbtx3x10jfrs0yq3qrfdx00000gn/T/ghc40705_0/ghc40705_0.s:3886:0: bad immediate value for offset (5788) /var/folders/qh/jdhbtx3x10jfrs0yq3qrfdx00000gn/T/ghc40705_0/ghc40705_0.s:3860:0: bad immediate value for offset (5700)

2 4

Fw: message
by Pam M 11 Mar '13

11 Mar '13

http://www.championsleaguefutbol.com/hufmbv/wkq/qt/emabu/mtixy/el/qaxqf.dufz

1 0

Has anyone submitted an app using GHC-IOS?
by Robert Lorentz 03 Mar '13

03 Mar '13

Curious if this would be approved in the App Store especially with respect to the warnings you get when linking. Has anyone succeeded? I would be willing to do this exercise with one of my apps (simply put in a rev that adds the hello world type example) - but if someone else has done it I won't waste my time

2 1