> Simon Marlow wrote:
> > [...] in fact if you check out the sources at the moment
> you'll get a
> > Haskell 98 compiler with a generational garbage collector :-)
>
> Hmmm, wishful thinking... If I check out at the moment I get:
I didn't say it was a *working* compiler :-)
> * a module ParseIface which needs 130MB (!) heap for compilation
> with ghc-4.01 on Linux
4.02 will help, but I'm still using 3.02 to bootstrap with.
> * a ghc which can't compile its own Prelude:
>
> ../../../ghc/driver/ghc -H100M -pgmc/usr/bin/gcc
> -optc-B/usr/lib/gcc-lib/i486-linux/2.7.2.3/ -recomp -cpp
> -fglasgow-exts -fvia-C -Rghc-timing -O -split-objs -odir
> PrelBase -H10m -c PrelBase.lhs -o PrelBase.o -osuf o
> ghc: ignoring heap-size-setting option (-H10m)...not
> the largest seen
> ghc: ignoring heap-size-setting option (-H10m)...not
> the largest seen
>
> PrelBase.lhs:18: Could not find valid interface file `PrelErr'
>
> PrelBase.lhs:19: Could not find valid interface file `PrelGHC'
I don't know what problem you're having here, but it looks suspiciously like
the libc5/libc6 conflict problems we were having. I solved those here by
compiling up a 3.02 from .hc files on our libc6 system.
> * a runtime system without the recent patches (Adjustor.c,
> Storage.h)
Still on my todo list, don't worry!
> * a ghc driver script which still confuses -M and -H
I plan to fix this but it's not as easy as it looks. We still support
bootstrapping the compiler with GHC 3.xx which has the old RTS and therefore
expects to see +RTS -H<wibble>, so the current driver script will have to
support both types of RTS. And the -M option is already used, sigh.
Cheers,
Simon
--
Simon Marlow
Microsoft Research Ltd., Cambridge, U.K.
