November 2005 - Glasgow-haskell-users

RE: [Haskell-cafe] GHC ignores RTS options
by Simon Marlow 16 Nov '05

16 Nov '05

On 16 November 2005 13:24, Tomasz Zielonka wrote: > On Wed, Nov 16, 2005 at 01:09:07PM -0000, Simon Marlow wrote: >>> BTW - it would be nice if it was easier to override the default RTS >>> options when compiling the program. Perhaps a nice task for a newbie >>> GHC developer? >> >> Already possible, see: >> >> http://www.haskell.org/ghc/docs/latest/html/users_guide/runtime-control. >> html#rts-hooks > > I was aware of this, and I should write about it. > >> Admittedly it's a bit more fiddly than a command-line flag to GHC, >> though. > > And because of this I wrote that it would be nice if it was easier :-) Sorry, I misread the message. > I've bitten several times by the default values of RTS options, > especially the stack size limit (-K). But it was always easier to > set GHCRTS or use +RTS -RTS, so I never used the rts-hooks. > > Do you think I could do this? I am not sure how difficult it would be. Sure, it shouldn't be too hard. You need to generate a .c file containing the ghc_rts_opts setting, and then compile & link it into the final binary. It's just a matter of plumbing the various bits together in GHC's front end. (it might be nicer to generate Cmm and then push this through the back end, to avoid the dependence on the C compiler, though). Cheers, Simon

2 1

RE: GHC 6.4.1 on Solaris x86
by Simon Marlow 16 Nov '05

16 Nov '05

On 16 November 2005 13:18, Duncan Coutts wrote: > On Wed, 2005-11-16 at 12:59 +0000, Simon Marlow wrote: >> On 14 November 2005 04:17, Ben Horsfall wrote: >> >>> Is there a binary package for GHC 6.4.1 on Solaris x86 (Sun OS 5.9)? >>> >>> Building from source fails thus: >>> >>> ../../ghc/compiler/ghc-inplace -optc-O -optc-Wall -optc-W >>> -optc-Wstrict-prototypes -optc-Wmissing-prototypes >>> -optc-Wmissing-declarations -optc-Winline -optc-Waggregate-return >>> -optc-Wbad-function-cast -optc-I../includes -optc-I. -optc-Iparallel >>> -optc-DCOMPILING_RTS -optc-fomit-frame-pointer -H16m -O -O2 -static >>> -I. -#include Prelude.h -#include Rts.h -#include RtsFlags.h >>> -#include RtsUtils.h -#include StgRun.h -#include Schedule.h >>> -#include Printer.h -#include Sanity.h -#include STM.h -#include >>> Storage.h -#include SchedAPI.h -#include Timer.h -#include Itimer.h >>> -#include ProfHeap.h >>> -#include LdvProfile.h -#include Profiling.h -#include Apply.h >>> -fvia-C >>> -dcmm-lint -c StgCRun.c -o StgCRun.o >>> Assembler: StgCRun.c >>> "/tmp/ghc14327.s", line 18 : Illegal mnemonic >>> "/tmp/ghc14327.s", line 18 : Syntax error >>> "/tmp/ghc14327.s", line 18 : Illegal mnemonic >>> make[2]: *** [StgCRun.o] Error 1 >> >> There isn't a binary for 6.4.1 that I know of. You should be able to >> follow the porting instructions in the Building Guide to get an >> unregisterised version going (get the version from CVS, there were >> some recent modifications). > > My department uses 6.4.1 on Solaris 9 and it works nicely. They built > it from source using version 6.2.2 (for which there are binaries I > believe). I don't think it's necessary to do a fresh port. Ah good. Any chance of building a binary dist I can put up? Cheers, Simon

2 1

RE: [Haskell-cafe] GHC ignores RTS options
by Simon Peyton-Jones 16 Nov '05

16 Nov '05

I'd love to see this done. It's been requested several times. Simon | -----Original Message----- | From: glasgow-haskell-users-bounces(a)haskell.org [mailto:glasgow-haskell-users- | bounces(a)haskell.org] On Behalf Of Simon Marlow | Sent: 16 November 2005 15:48 | To: Tomasz Zielonka | Cc: Dusan Kolar; GHC-users; Haskell-Cafe(a)haskell.org | Subject: RE: [Haskell-cafe] GHC ignores RTS options | | On 16 November 2005 13:24, Tomasz Zielonka wrote: | | > On Wed, Nov 16, 2005 at 01:09:07PM -0000, Simon Marlow wrote: | >>> BTW - it would be nice if it was easier to override the default RTS | >>> options when compiling the program. Perhaps a nice task for a newbie | >>> GHC developer? | >> | >> Already possible, see: | >> | >> | http://www.haskell.org/ghc/docs/latest/html/users_guide/runtime-control. | >> html#rts-hooks | > | > I was aware of this, and I should write about it. | > | >> Admittedly it's a bit more fiddly than a command-line flag to GHC, | >> though. | > | > And because of this I wrote that it would be nice if it was easier :-) | | Sorry, I misread the message. | | > I've bitten several times by the default values of RTS options, | > especially the stack size limit (-K). But it was always easier to | > set GHCRTS or use +RTS -RTS, so I never used the rts-hooks. | > | > Do you think I could do this? I am not sure how difficult it would be. | | Sure, it shouldn't be too hard. You need to generate a .c file | containing the ghc_rts_opts setting, and then compile & link it into the | final binary. It's just a matter of plumbing the various bits together | in GHC's front end. | | (it might be nicer to generate Cmm and then push this through the back | end, to avoid the dependence on the C compiler, though). | | Cheers, | Simon | _______________________________________________ | Glasgow-haskell-users mailing list | Glasgow-haskell-users(a)haskell.org | http://www.haskell.org/mailman/listinfo/glasgow-haskell-users

1 0

RE: [Haskell-cafe] GHC ignores RTS options
by Simon Marlow 16 Nov '05

16 Nov '05

On 16 November 2005 08:12, Tomasz Zielonka wrote: > On Wed, Nov 16, 2005 at 08:53:18AM +0100, Dusan Kolar wrote: >> I wonder whether anybody could advise me how to pass >> correctly RTS options to ghc? I've tried > > You have to pass these options to the program, not > to GHC. > > BTW - it would be nice if it was easier to override the default RTS > options when compiling the program. Perhaps a nice task for a newbie > GHC developer? Already possible, see: http://www.haskell.org/ghc/docs/latest/html/users_guide/runtime-control. html#rts-hooks Admittedly it's a bit more fiddly than a command-line flag to GHC, though. Cheers, Simon

2 1

RE: GHC 6.4.1 on Solaris x86
by Simon Marlow 16 Nov '05

16 Nov '05

On 14 November 2005 04:17, Ben Horsfall wrote: > Is there a binary package for GHC 6.4.1 on Solaris x86 (Sun OS 5.9)? > > Building from source fails thus: > > ../../ghc/compiler/ghc-inplace -optc-O -optc-Wall -optc-W > -optc-Wstrict-prototypes -optc-Wmissing-prototypes > -optc-Wmissing-declarations -optc-Winline -optc-Waggregate-return > -optc-Wbad-function-cast -optc-I../includes -optc-I. -optc-Iparallel > -optc-DCOMPILING_RTS -optc-fomit-frame-pointer -H16m -O -O2 -static > -I. -#include Prelude.h -#include Rts.h -#include RtsFlags.h -#include > RtsUtils.h -#include StgRun.h -#include Schedule.h -#include Printer.h > -#include Sanity.h -#include STM.h -#include Storage.h -#include > SchedAPI.h -#include Timer.h -#include Itimer.h -#include ProfHeap.h > -#include LdvProfile.h -#include Profiling.h -#include Apply.h -fvia-C > -dcmm-lint -c StgCRun.c -o StgCRun.o > Assembler: StgCRun.c > "/tmp/ghc14327.s", line 18 : Illegal mnemonic > "/tmp/ghc14327.s", line 18 : Syntax error > "/tmp/ghc14327.s", line 18 : Illegal mnemonic > make[2]: *** [StgCRun.o] Error 1 There isn't a binary for 6.4.1 that I know of. You should be able to follow the porting instructions in the Building Guide to get an unregisterised version going (get the version from CVS, there were some recent modifications). Cheers, Simon

2 1

RE: heap profiling and name truncation
by Simon Marlow 16 Nov '05

16 Nov '05

On 14 November 2005 20:35, Abraham Egnor wrote: > I'm trying to use the built-in ghc heap profiling and running into a > large roadblock: the names of producers are truncated, often into > uselessness, and there doesn't seem to be an option to control this. > Am I missing something in the docs? The limit of 25 characters before the "..." is built-in, I'm afraid. How ever, the cost centre ID is included (the number in parentheses before the cost centre stack), and you can match this to the full CCS by adding the -P option: $ ./prog +RTS -hc -P also produces a time profile in prog.prof that includes the ID of each cost centre stack. Cheers, Simon

1 0

RE: working lazy memoizer for GHC?
by Simon Marlow 16 Nov '05

16 Nov '05

The whole util package is deprecated, because it is using the old non-hierarchical module names. The memo table library should really be re-packaged with Cabal and made available somewhere. Cheers, Simon On 16 November 2005 07:29, Conal Elliott wrote: > I'd also like to know why this module is deprecated. Does anyone > know? - Conal > > -----Original Message----- > From: glasgow-haskell-users-bounces(a)haskell.org > [mailto:glasgow-haskell-users-bounces@haskell.org] On Behalf Of > Chung-chieh Shan > Sent: Monday, November 14, 2005 5:46 PM > To: glasgow-haskell-users(a)haskell.org > Subject: Re: working lazy memoizer for GHC? > > Conal Elliott <conal(a)conal.net> wrote in article > <011c01c5e655$97329390$0202a8c0@conalputer> in > gmane.comp.lang.haskell.glasgow.user: >> Does anyone have a working lazy memoizer (memo :: (a->b) -> (a->b)) >> for GHC? - Conal > > There's a deprecated one in the util package. Why is it deprecated > anyway?

1 0

Re: [Haskell-cafe] GHC ignores RTS options
by Tomasz Zielonka 16 Nov '05

16 Nov '05

On Wed, Nov 16, 2005 at 08:53:18AM +0100, Dusan Kolar wrote: > I wonder whether anybody could advise me how to pass > correctly RTS options to ghc? I've tried You have to pass these options to the program, not to GHC. BTW - it would be nice if it was easier to override the default RTS options when compiling the program. Perhaps a nice task for a newbie GHC developer? Best regards Tomasz

1 0

working lazy memoizer for GHC?
by Conal Elliott 16 Nov '05

16 Nov '05

Does anyone have a working lazy memoizer (memo :: (a->b) -> (a->b)) for GHC? - Conal

2 2

heap profiling and name truncation
by Abraham Egnor 14 Nov '05

14 Nov '05

I'm trying to use the built-in ghc heap profiling and running into a large roadblock: the names of producers are truncated, often into uselessness, and there doesn't seem to be an option to control this. Am I missing something in the docs? Abe

1 0