2.08 little bug
Just a minor thing in 2.08, install copies the driver to bin/ghc and sets a symlink from ghc-2.08 to ghc rather than the other way round so it overwrites the old copy... 94 -rwxr-xr-x 1 jpc195903 Oct 9 17:56 ghc 1 lrwxrwxrwx 1 jpc13 Oct 2 17:01 ghc-2.07 - ghc 1 lrwxrwxrwx 1 jpc13 Oct 9 17:56 ghc-2.08 - ghc totally painless install (from sparc solaris binary) otherwise. justin
Re: pattern match
I am using GHC 2.06 for Win32 and why does GHC complain about "possibly incomplete patterns"? It does so because it uses a brain-dead and incorrect way to detect incomplete patterns. (I can say this because I did it myself.) As we speak Juan Quintela is doing a Better Job. It'll appear soon. Thanks for the report. I'm keen to know about obscure or (as in this case) incorrect error messages. Simon
Another problem recompiling 2.07 with itself
This is not really a bug in ghc itself, rather a problem with its makefile system: While trying to recompile 2.07 with itself I encountered the following error: ==fptools== make all; in /.automount/kawa/export/linux/ben/Dlubanie/Build/fptools/ghc/lib rm -f ghc/PrelBase.o ; if [ ! -d ghc/PrelBase ]; then mkdir ghc/PrelBase; else find ghc/PrelBase -name '*.o' -print | xargs rm -f __rm_food ; fi ../../ghc/driver/ghc -recomp -cpp -fglasgow-exts -fvia-C -Rghc-timing -O -split-objs -odir ghc/PrelBase -H12m -c ghc/PrelBase.lhs -o ghc/PrelBase.o -osuf o Can't locate ghc-split.prl in @INC at ../../ghc/driver/ghc line 2938. Regards, Marcin Benke
Re: Another problem recompiling 2.07 with itself
Marcin Benke [EMAIL PROTECTED] writes: ==fptools== make all; in /.automount/kawa/export/linux/ben/Dlubanie/Build/fptools/ghc/lib rm -f ghc/PrelBase.o ; if [ ! -d ghc/PrelBase ]; then mkdir ghc/PrelBase; else find ghc/PrelBase -name '*.o' -print | xargs rm -f __rm_food ; fi ../../ghc/driver/ghc -recomp -cpp -fglasgow-exts -fvia-C -Rghc-timing -O -split-objs -odir ghc/PrelBase -H12m -c ghc/PrelBase.lhs -o ghc/PrelBase.o -osuf o Can't locate ghc-split.prl in @INC at ../../ghc/driver/ghc line 2938. The build system has failed to find the canonical directory in which your source tree lives, probably due to the automounter business. If you look in configure.in, you'll find some hacks we've put in the there that solve the problem at Glasgow, but there doesn't seem to be a general solution. Workaround: set FPTOOLS_TOP_ABS explicitly in build.mk. -- Simon Marlow [EMAIL PROTECTED] University of Glasgow http://www.dcs.gla.ac.uk/~simonm/ finger for PGP public key
UnHappy
Dear Buggies, `make all' is not successful (booting with ghc-0.29) because there are some Happy generated files containing `newtype' declarations which ghc-0.29 does not swallow. Diagnostics: ghc-0.29 -cpp -fglasgow-exts -Rghc-timing -I. -IcodeGen -InativeGen -Iparser -iutils:basicTypes:types:hsSyn:prelude:rename:typecheck:deSugar:coreSyn:specialise:simplCore:stranal:stgSyn:simplStg:codeGen:absCSyn:main:reader:profiling:parser:nativeGen -fhaskell-1.3 -fomit-derived-read -fomit-reexported-instances -Onot -H16m -H48M -c rename/ParseType.hs -o rename/ParseType.o -osuf o "rename/ParseType.hs", line 727, column 15: parse error on input: "-" The following files are concerned: rename/ParseType.hs rename/ParseUnfolding.hs rename/ParseIface.hs each containing newtype HappyState b c = HappyState (Int# --- token number Int# --- token number (yes, again) b - -- token semantic value HappyState b c - -- current state [HappyState b c] --- state stack c) Renaming `newtype' to `data' seems to work ... [the workstation is still busy] Ralf
Re: Compiler bug pops up when compiling ghc-2.07 with 2.02
Simon L Peyton Jones writes: I tried, but it seems to complete successfully. Nevertheless, later make all fails with the same message. Probably the dependencies generated by make boot (or make depend, for that matter) _are_ circular (citing from ghc/compiler/.depend): utils/FastString.o : basicTypes/Unique.hi basicTypes/Unique.o : utils/Pretty.hi utils/Pretty.o : utils/FastString.hi My guess is that you have not got the line Ghc2_0= YES in your build.mk file. GHC 2.x uses a different way of dealing with mutual recursion (via M.hi-boot files) than GHC 0.29 (which used M.lhi files). No, I had this line in my build.mk from the beginning. For the record, here goes my builld.mk: GhcBuilderVersion = 207 WithGhcHc = ghc-2.07 Ghc2_0=YES GhcWithHscBuiltViaC=NO Remember, don't bother to use 2.06 or earlier for self-booting. Only 2.07 is up to it. Yes, I use 2.07 (or so it seems): [08:53:35] ben@kawa:~ ghc --version The Glorious Glasgow Haskell Compilation System, version 2.07,patchlevel 0 BTW, when you are so kind as to answer my not-so-wise questions, what is the 'proper' way of getting a .dvi from literate sources? I could not find anything in the docs, and my attempts to follow the literate/README fail at the make stage already. Regards, Marcin Benke "Calm down, it's only ones and zeros!" - DJ Delorie
Installing 2.08
Installing ghc-2.08 on Linux and HPUX went quite smoothly. Two minor bugs: * Some $'s disappeared from Perl scripts (e.g. ghc, hscpp, ...). This has been reported earlier (incl. fix) and seems to happen with Bash 2.01 or later (which is quite unfortunate as Bash 2.01 *is* /bin/sh on newer Linux systems). Example: datadir=/foo/bar instead of $datadir=/foo/bar * During "make binary-dist" lit2texi stumbles three times like this: [...] Making html documentation in ghc/docs/users_guide ../../../literate/lit2texi -c-o prof-compiler-options.itxi prof-compiler-options.lit lit2texi: error(s) from prefilter pipe make[1]: *** [prof-compiler-options.itxi] Error 1 Consequently, some docs are missing later. Snapshots at the usual place. When binary-dist is working, I'll switch to normal Glaswegian tar files instead of snapshots. -- Sven PanneTel.: +49/89/2178-2235 LMU, Institut fuer Informatik FAX : +49/89/2178-2211 LFE Programmier- und Modellierungssprachen Oettingenstr. 67 mailto:[EMAIL PROTECTED]D-80538 Muenchen http://www.pms.informatik.uni-muenchen.de/mitarbeiter/panne