cvs commit: fptools/ghc/compiler/basicTypes NewDemand.lhs IdInfo.lhs MkId.lhs fptools/ghc/compiler/coreSyn PprCore.lhs fptools/ghc/compiler/main DriverState.hs fptools/ghc/compiler/simplCore OccurAnal.lhs SimplCore.lhs fptoo

[Simon Peyton-Jones]

I'm still debugging it!  Stay tuned

| -----Original Message-----
| From: Marcin 'Qrczak' Kowalczyk [mailto:qrczak@knm.org.pl]=20
| Sent: 18 July 2001 20:54
| To: cvs-ghc@haskell.org
| Subject: Re: cvs commit: fptools/ghc/compiler/basicTypes=20
| NewDemand.lhs IdInfo.lhs MkId.lhs=20
| fptools/ghc/compiler/coreSyn PprCore.lhs=20
| fptools/ghc/compiler/main DriverState.hs=20
| fptools/ghc/compiler/simplCore OccurAnal.lhs SimplCore.lhs fptoo
|=20
|=20
| Tue, 17 Jul 2001 08:28:31 -0700, Simon Peyton Jones=20
| <simonpj@glass.cse.ogi.edu> pisze:
|=20
| >   	If you build the compiler with -DDEBUG,
| >   	you'll get the demand analyser, but the existing
| >   	strictness analyser etc are still there.  All the
| >   	demand analyser does is to compare its output with
| >   	the existing stuff and report differences.
|=20
| Results of the analysis are strange, or I don't understand them.
|=20
| For example here
|     WORSE strictness for Util.zipWith4Equal Old LLVVVVT
|                                             New LLVVLLT
|     WORSE strictness for Util.seqList Old VVT
|                                       New VLT
| the old result seems to be correct, the new one is too=20
| pessimistic. OTOH here
|     BETTER strictness for Pretty.fullRender Old LLLLLLT
|                                             New VLLLLVT
| the new one is probably correct, the old one is unnecessarily=20
| pessimistic.
|=20
| I didn't expect so many differences!
|=20
| --=20
|  __("<  Marcin Kowalczyk * qrczak@knm.org.pl=20
| http://qrczak.ids.net.pl/  \__/
|   ^^                     =20
| SYGNATURA ZAST=CAPCZA
| QRCZAK
|=20
|=20
| _______________________________________________
| Cvs-ghc mailing list
| Cvs-ghc@haskell.org http://www.haskell.org/mailman/listinfo/cvs-ghc
|=20
|=20