Re: heap and other problems self-compiling 4.02
Giuliano wrote: On Wed, May 19, 1999 at 10:16:54AM +0200, Thomas Pasch wrote: I tried to compile 4.02 from sources on a ix86-linux-glibc2.1 machine and found that this is simply impossible. But when it cames to compile ParselFace.hs in ghc/compiler/rename the Heap was exhausted. One solution is to add "GhcHcOpts = -optCrts-M100m" to mk/build.mk This is probably rather excessive, since the option will be set for every single file in the compiler. My build.mk file contains the lines: # Magic for the obscene amount of memory we need to compile the parser. rename/ParseIface_HC_OPTS += -H80m -optCrts-M90m which seems to do the trick, on sparc-solaris anyway. Perhaps they should be added to the GHC distribution?
RE: heap and other problems self-compiling 4.02
My build.mk file contains the lines: # Magic for the obscene amount of memory we need to compile the parser. rename/ParseIface_HC_OPTS += -H80m -optCrts-M90m which seems to do the trick, on sparc-solaris anyway. Perhaps they should be added to the GHC distribution? Well, the fact that we have to keep adding this helps to remind us that there's a bug to be fixed (plus the fact it's in my ToDo list and on the whiteboard in fairly big letters :-) Anyway, adding -dcore-lint is a generally good idea, and if you use the latest happy from the CVS repository it has some NOINLINE pragmas which bring the residency when compiling ParseIface.hs down to a perfectly respectable 16M or thereabouts. Now, the happy parser I'm working on for GHC is a different matter... Cheers, Simon
Re: heap and other problems self-compiling 4.02
Thomas Pasch wrote: [...] But when it cames to compile ParselFace.hs in ghc/compiler/rename the Heap was exhausted. IIRC, -dcore-lint radically cuts down the heap required for the compilation of this module. Cheers, Sven -- 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
RE: heap and other problems self-compiling 4.02
IIRC, -dcore-lint radically cuts down the heap required for the compilation of this module. And try -funfolding-use-threshold0 too. Cheers, Simon
Re: heap and other problems self-compiling 4.02
Hello. On Wed, May 19, 1999 at 10:16:54AM +0200, Thomas Pasch wrote: I tried to compile 4.02 from sources on a ix86-linux-glibc2.1 machine and found that this is simply impossible. But when it cames to compile ParselFace.hs in ghc/compiler/rename the Heap was exhausted. One solution is to add "GhcHcOpts = -optCrts-M100m" to mk/build.mk I downloaded the current source code. Ungzipping is possible with no errors but taring afterwards leads to a checksum error. Did you try tar xzf foo.tar.gz ? I have had no problems with unpacking the source archives. This suggests old/buggy tar or gzip (unlikely) or some corruption of the archive somehow. Giuliano -- [EMAIL PROTECTED]
Re: heap and other problems self-compiling 4.02
In my previous message I suggested "GhcHcOpts = foo", this should have been "+=". Giuliano -- [EMAIL PROTECTED]
RE: heap and other problems self-compiling 4.02
Isn't it about time a note about -dcore-lint was placed on http://research.microsoft.com/users/t-simonm/ghc/download_ghc_402.html Ok, Ok. Done. Simon