- jhc - Haskell.org

darcs patch: Upgrade some JHC_EXT_LIBS.
by kiwamu＠debian.or.jp 18 Mar '13

18 Mar '13

1 patch for repository http://repetae.net/repos/jhc: Sun Mar 17 16:06:29 JST 2013 kiwamu(a)debian.or.jp * Upgrade some JHC_EXT_LIBS. I tested compile test only for the patch. Are there any things that I should test something else? -- Kiwamu Okabe

2 1

Re: [jhc] [Haskell-cafe] ANNOUNCE: Ajhc 0.8.0.2 Release
by Kiwamu Okabe 17 Mar '13

17 Mar '13

Hi, On Sun, Mar 17, 2013 at 4:45 AM, Joachim Breitner <mail(a)joachim-breitner.de> wrote: > Great. Would it be possible for you two to agree on a mode that allows > Kiwamu to develop in his pace, but still avoid having a proper fork of > jhc pushed to distros and users? Something like: Kiwamu hacks away > happily on the ajhc repo and people interested in the very latest state > can pull from there, but ajhc releases get (if there are no technical > issues) merged into jhc, and jhc is the version that regular users > should use (and distributions should package). I think good, because I maintain jhs as Debian package already. http://ftp-master.debian.org/new/jhc_0.8.0~darcs20120314-1.html Maintaining Ajhc Debian package is bad idea, if jhc has the most functions had by Ajhc. How about below policy between jhc and Ajhc. 1. Jhc and Ajhc share same command name "jhc". 2. Jhc controls major version number. Example, jhc controls X, Y and Z in version "X.Y.Z". 3. Ajhc controls minor version number. Example, Ajhc controls only Q in version "X.Y.Z.Q". 4. Ajhc show below message with --version option. (A)jhc 0.8.0.1 (80aa12fb9b57622bba2f0e911d7ebc0c04ddb662) compiled by ghc-7.4 on a x86_64 running linux The policy is to keep minimum change between jhc and Ajhc. And I think that Ajhc having own command name confuses jhc user. Best regards, -- Kiwamu Okabe

1 0

darcs patch: Explain RTS cflags for tiny cpu.
by kiwamu＠debian.or.jp 16 Mar '13

16 Mar '13

1 patch for repository http://repetae.net/repos/jhc: Sun Mar 17 02:21:07 JST 2013 kiwamu(a)debian.or.jp * Explain RTS cflags for tiny cpu.

2 1

[patch] New RTS for tiny CPU
by Kiwamu Okabe 16 Mar '13

16 Mar '13

Hi John. I think the patch is useful to run Haskell code on tiny CPU and small memory size. Github link is below. https://github.com/ajhc/ajhc/commit/3167551530b0576cf1f42f928865868ce9aa0b50 The changes are... 1. Control BLOCK_SIZE and MEGABLOCK_SIZE from Makefile. 2. Do not allocate MEGABLOCK dynamically. 3. Set function pointer's bit0 for Cortex-M3. http://communities.mentor.com/community/cs/archives/arm-gnu/msg01904.html 4. Naive GC flag to run GC when no blocks. 5. Control stack_grow size. Example of usage is found at https://github.com/ajhc/demo-cortex-m3. Best regards, -- Kiwamu Okabe

2 3

ANNOUNCE: Ajhc 0.8.0.2 Release
by Kiwamu Okabe 16 Mar '13

16 Mar '13

We are happy to announce Ajhc 0.8.0.2. It's first release announce for Ajhc. Major change on this release is ability to compile Haskell code for tiny CPU. There is demo on tiny CPU at https://github.com/ajhc/demo-cortex-m3. And you can watch demo movie at http://www.youtube.com/watch?v=bKp-FC0aeFE. Perhaps changes on the announce will be merged to jhc. Ajhc's project web site is found at http://ajhc.masterq.net/. You can get Ajhc 0.8.0.2 source code from https://github.com/ajhc/ajhc/tags. ## Changes * Fix warning messages on compiling. * Ready to compile with GHC 7.6.2. * New RTS for tiny CPU. How to use: https://github.com/ajhc/demo-cortex-m3#porting-the-demo-to-a-new-platform - - - Metasepi team

2 1

Re: [jhc] [patch] New RTS for tiny CPU
by jorge.pelizzoni＠gmail.com 16 Mar '13

16 Mar '13

2 1

[patch] Fix warning messages when compiling.
by Kiwamu Okabe 16 Mar '13

16 Mar '13

1 0

[patch] Ready to compile with GHC 7.6.2
by Kiwamu Okabe 16 Mar '13

16 Mar '13

Hi John. I think jhc with attached patch can be compiled on GHC 7.6.2. Could you review it? Github links are below. https://github.com/ajhc/ajhc/commit/a187199849a4a4cef21bf24c87548374673d81e5 https://github.com/ajhc/ajhc/commit/8b41239136cc607eed0d7974b0d6eb643b95faf8 Best regards, -- Kiwamu Okabe

1 0

Re: [jhc] [BUG] Poke should use volatile.
by Kiwamu Okabe 07 Mar '13

07 Mar '13

Hi Henning. I don't know detail of C language... On Thu, Mar 7, 2013 at 9:07 PM, Henning Thielemann <lemming(a)henning-thielemann.de> wrote: > Yes - at least this is the expected semantics of the IO monad. > Unless someone has written a RULE that removes the duplicate peek. However, > I think this is generally not a good idea, since even regular memory can be > modified by a concurrent thread. So clear. Thank's a lot. -- Kiwamu Okabe

1 0

[BUG] Poke should use volatile.
by Kiwamu Okabe 07 Mar '13

07 Mar '13

Hi. I think poke function should use volatile on C language side. If not, the code created with jhc cannot write raw memory. diff --git a/src/C/FromGrin2.hs b/src/C/FromGrin2.hs index 46e1a08..d4508bd 100644 --- a/src/C/FromGrin2.hs +++ b/src/C/FromGrin2.hs @@ -712,7 +712,7 @@ convertPrim p vs ty | (Poke t) <- p, [v,x] <- vs = do v' <- convertVal v x' <- convertVal x - return $ expressionRaw ("*((" <> (opTyToC' t) <+> "*)" <> (parens $ renderG v') <> text ") = " <> renderG x') + return $ expressionRaw ("*((volatile " <> (opTyToC' t) <+> "*)" <> (parens $ renderG v') <> text ") = " <> renderG x') | (AddrOf _ t) <- p, [] <- vs = do rt <- convertTypes ty return . cast rt $ expressionRaw ('&':unpackPS t) I found it when use jhc to program Cortex-M3. https://gitorious.org/metasepi/demo-cortex-m3 Regards, -- Kiwamu Okabe

2 11