October 2006 - Yhc - Haskell.org

Yhc HCAR Entry
by Neil Mitchell 25 Oct '06

25 Oct '06

Hi, Last year we were in the HCAR: http://haskell.org/communities/06-2006/html/report.html#yhc Quote: The York Haskell Compiler (yhc) is a backend rewrite of the nhc98 (→2.3) compiler to support features such as a platform independent bytecode and runtime system. It is currently work in progress, it compiles and correctly runs almost every standard Haskell 98 program but FFI support is on going. Contributions are welcome. We need a better description of what we do, what we've done and where we are going, since that contribution is a massive undersell! Anyone want to put together something? Anyone have anything that needs to be put in? Thanks Neil

1 0

Yhc Core breakage
by Neil Mitchell 25 Oct '06

25 Oct '06

Hi, I just change Yhc.Core in an imcompatible way, and upgraded to version 2 of the files. The new difference is that Functions and Variables are treated differently, which makes some things easier for the users of Core. Any existing .ycr files will fail to load, so delete them, rebuild them and then use them. I also added a -viewcore argument to Yhc today, yhc -viewcore File.ycr will now show you the file. Woohoo! Thanks Neil

1 0

Autogenerate .ycr for each .hbc?
by Dimitry Golubovsky 24 Oct '06

24 Oct '06

Hi, I need to have a .ycr (YHC core) file generated for each .hbc file that is generated when building YHC (prelude, packages). I am guessing, I have to add the -corep option somewhere in Scons building scripts. Can this be done in a single place (toplevel SConstruct), or in multiple? In general it would be nice to make core autobuilding as an option, i. e. when "core" appears in the scons command line then generate binary core files along with bytecode files (by appending that -corep option where appropriate). PS I am not very much familiar with Python and Scons. I tried hard to see where YHC commandline flags are set to generate bytecodes for prelude and packages, but couldn't find that out. Thanks for any advice. -- Dimitry Golubovsky Anywhere on the Web

2 2

Fwd: [York Haskell Compiler] 10/23/2006 01:02:04 AM
by Neil Mitchell 23 Oct '06

23 Oct '06

A comment left on blogger, which probably should be more widely circulated. Not sure if this will be useful to us or not, but I'll let the Linux people decide. If it looks useful/helpful/a good idea can someone open a bug on it? -------------------------- Hi - I was just wondering - for Linux installs, have you looked at doing an "autopackage" for YHC? ( http://autopackage.org/ ) That is a distro-neutral format which is very easy to install, and using it could save a lot of time doing packages for each distro. There is even a related utility called "Packer" http://packer.sourceforge.net/ From its home-page - "Packages are a great convenience for gurus and new users alike. Thus, users may be less inclined to use software that is not packaged. This is unfortunate for GNU/Linux software developers since packaging software is a huge pain, especially given the huge variety of packaging formats. Packer helps to automate this process. Using a text-based wizard, packer can collect all the information regarding a program needed to create a package. Then, a simple command is all it takes to create a debian directory (used to build debian packages), RPM SPEC files (for Mandriva, SuSE, and Fedora/Red Hat), and autopackages spec files (autopackage is a system for creating distribution-neutral installers with dependency handling). These files are of a similar quality to hand-crafted ones. Packer can also automatically build the packages themselves if requested." Just thought I'd mention those two things - keep up the great work! :-) - Andy -- Posted by Anonymous to York Haskell Compiler at 10/23/2006 01:02:04 AM

1 0

lexical error in interactive/Interact.hs building Yhc
by Andrew Wilcox 22 Oct '06

22 Oct '06

$ scons [ ... ] Getting dependencies for src/interactive/CmdLine.hs. An error ocurred running `ghc -M -optdep-f -optdepsrc/interactive/.CmdLine.hs.dep -fwarn-missing-signatures -i./src/compiler98:depends/cpphs:depends/filepath:./src/interactive src/interactive/CmdLine.hs`: > >./src/interactive/Interact.hs:17:1: lexical error > Dependencies will not be calculated. ghc -i./src/compiler98 -idepends/cpphs -idepends/filepath -i./src/interactive -fwarn-missing-signatures -c src/interactive/CmdLine.hs -o src/interactive/CmdLine.o src/interactive/CmdLine.hs:3:0: Failed to load interface for `Interact': Could not find module `Interact': use -v to see a list of the files searched for scons: *** [src/interactive/CmdLine.o] Error 1 scons: building terminated because of errors. Perhaps ghc needs the -cpp flag? $ darcs changes --last 5 Sat Oct 21 13:52:49 EDT 2006 Andrew Wilkinson <aw(a)cs.york.ac.uk> * Fix yhc build error on Windows buildbot. Sat Oct 21 13:21:16 EDT 2006 Andrew Wilkinson <aw(a)cs.york.ac.uk> * Python people have moved the directory for libffi. Sat Oct 21 13:04:42 EDT 2006 Andrew Wilkinson <aw(a)cs.york.ac.uk> * Fix building the docs now yhe can be built. Sat Oct 21 10:30:44 EDT 2006 Andrew Wilkinson <aw(a)cs.york.ac.uk> * Add support for building Yhe. Fri Oct 20 09:50:37 EDT 2006 Andrew Wilkinson <aw(a)cs.york.ac.uk> * Fix building the docs. $ ghc -V The Glorious Glasgow Haskell Compilation System, version 6.4.2 $ scons configure scons: Reading SConscript files ... Building version 0.7.0-33-custom. Checking for architecture... x86 Checking for operating system... linux2 Checking for C header file gmp.h... yes Checking for mpz_t integ; mpz_init (integ) in C library gmp... yes Checking for libgmp version... 4.2.1 Checking for Python version 2.3 or later... Found version 2.4.3 final (0). Checking for Scons binary... "/usr/bin/scons" Checking for Scons version... Found version 0.96.1.D001. Checking for Darcs binary... "/usr/bin/darcs" Checking for Darcs version... Found version 1.0.8 (release). Checking for Subversion binary... "/usr/bin/svn" Checking for Subversion version... Found version 1.3.2. Checking for GHC binary... "/opt/ghc/bin/ghc" Checking for GHC version 6.4.1 or later... Found version 6.4.2. Checking whether byte ordering is bigendian... no Checking size of char... 1 Checking size of short... 2 Checking size of int... 4 Checking size of long... 4 Checking size of long long... 8 Checking size of float... 4 Checking size of double... 8 Checking size of void*... 4 Checking for pthread_create() in C library pthread... yes Generating config.h with the following settings: WORDS_BIGENDIAN False HAVE_LIBPTHREAD 1 SIZEOF_LONG_LONG 8 SIZEOF_LONG 4 WIN32 0 SIZEOF_VOIDP 4 SIZEOF_CHAR 1 SIZEOF_DOUBLE 8 SIZEOF_FLOAT 4 SIZEOF_INT 4 HAVE_GCC_LABELS 1 SIZEOF_SHORT 2 scons: done reading SConscript files. scons: Building targets ... scons: Nothing to be done for `configure'. scons: done building targets. Thanks, Andrew

3 3

Yhc Version number
by Andrew Wilkinson 21 Oct '06

21 Oct '06

Hi, At the request of Neil I'm adding version numbering to the Yhc build process. What version number should we start at? Currently Yhi says it's version 1.0 while Yhc says 'yhc version XP'. Both of these seem a little odd to me - I have no idea what version XP means, and are we really at 1.0 state? Andrew

4 8

Problem building Yhe
by Andrew Wilkinson 21 Oct '06

21 Oct '06

I've written the scons scripts to build Yhe, but it currently fails to build for me. Can anyone explain this error message to me? ghc -i./src/compiler98 -idepends/cpphs -idepends/filepath -i./src/interactive -cpp -c src/interactive/GuiGtk.hs -o src/interactive/GuiGtk.o src/interactive/GuiGtk.hs:395:26: Couldn't match `TagName' against `Maybe a' Expected type: TagName Inferred type: Maybe a In the application `Just "input"' In the first argument of `textTagNew', namely `(Just "input")' src/interactive/GuiGtk.hs:845:24: Couldn't match `TagName' against `Maybe a' Expected type: TagName Inferred type: Maybe a In the first argument of `textTagNew', namely `Nothing' In a 'do' expression: linktag <- textTagNew Nothing I have gtk2hs-0.9.10 installed. Andrew

3 3

Issues building yhc
by Dimitry Golubovsky 16 Oct '06

16 Oct '06

Hi, I have tried (successfully this time) to build yhc, and encountered some issues: 1. Is there any particluar reason that "darcs get --partial" is done on cpphs? This results in Language and Text module hierarchies not pulled, which further results in compile error. I have made the following changes: ========================= bash$ darcs whatsnew { hunk ./Makefile 12 - darcs get --partial --repo-name=depends/cpphs $(CPPHS_URL) + darcs get --repo-name=depends/cpphs $(CPPHS_URL) hunk ./build/depends.py 12 - env.Execute(env.Action("darcs get --partial --repo-name=depends/cpphs http://www.cs.york.ac.uk/fp/darcs/cpphs")) + env.Execute(env.Action("darcs get --repo-name=depends/cpphs http://www.cs.york.ac.uk/fp/darcs/cpphs")) } ========================= basically removed "--partial" 2. I had to add a line to depends/ctypes/libffi/include/ffi.h: #define __LONG_LONG_MAX__ 9223372036854775807LL before #ifdef LONG_LONG_MAX # define FFI_LONG_LONG_MAX LONG_LONG_MAX #else . . . This may be a problem with my C compiler, because its own "limits.h" has definition for __LONG__LONG_MAX, but for some reason it is not visible to the compiler. So this is not a suggested fix, but just a question whether anybody had to do this for an old gcc (egcs 2.91.66 in my case)? Without that line, I got an error about unsupported 64 bit types (although the compiler supports long longs perfectly). 3. This may be a question to the scons team, but since it is used for this particukar project, I'm asking here first. How can I specify a custom path to a library, include file, etc.? With autoconf, I am usually able to specify "--with-bla=/foo/bar" provided that the developer of configure.ac made such an option available. How can the same be done with scons? E. g. I want to use alternate libgmp instead of one installed in /usr/lib? 4. Indeed, gmp 3.1 is not good: it is missing gmp_ printf. gmp 3.3 does have it. -- Dimitry Golubovsky Anywhere on the Web

3 2

Warning: scons fullclean recommended
by Dimitry Golubovsky 15 Oct '06

15 Oct '06

Hi, Samuel Bronson wrote: > I'm not sure I like that fix much... since when are Ids supposed to > resemble numbers? I' d vote for Neil's fix. It would be helpful to people (inclufing myself) converting binary Core into some other language: "v355" looks much better and much closer to a variable identifier than "vId 355". Thanks -- Dimitry Golubovsky Anywhere on the Web

1 0

examining yhc bytecode
by Kartik Vaddadi 14 Oct '06

14 Oct '06

Hello, I want to examine yhc bytecode, but Robert Dockins's library (http://www.eecs.tufts.edu/~rdocki01/yhc-bytecode.html) complains of a "Bad version". Are there any alternatives, other than doing it manually using a hex editor? Thanks. -- Kartik Vaddadi. Home: www.cse.iitb.ac.in/~kart Blogs: kartik-log.blogspot.com, kartik-rlog.blogspot.com Alternate mail ID: kartik.vad(a)gmail.com "50% Reservation, 100% Politics" - Protest the Indian government's decision to increase reservation in private educational institutions (yfemumbai.blogspot.com)

4 9