- Glasgow-haskell-users

RE: Complexity bug in garbage collector?
by Simon Marlow 29 Jul '05

29 Jul '05

On 29 July 2005 12:53, Josef Svenningsson wrote: > My machine was definitely swapping when I aborted the program. Can that > have any effect on the timings? yes, there's your problem. It's still a nice regular-looking curve though, which is mildly surprising. Simon

1 0

RE: Complexity bug in garbage collector?
by Simon Marlow 29 Jul '05

29 Jul '05

On 16 April 2005 23:19, Josef Svenningsson wrote: > OK, I've cooked up this little program to study the behaviour a > little closer: \begin{code} > module Main where > > main = print $ strictId [1..] > > strictId list = let (c,c') = work list c' > in c > where work [] y' = (y',[]) > work (x:xs) y' = (v,x:v') > where (v,v') = work xs y' > \end{code} > > This program just allocates like crazy til it dies. The funny looking > strictId function is just the strict identity function on lists. (Yes, > there are simpler ways to achieve the same thing. I just think the > above function is particularly sweet :-) > > I do the following: > $ ghc -prof -auto-all --make Main.hs > $ main.exe +RTS -hd -M<VERY MUCH> > > The resulting graph is suspiciously similar in shape to the one of my > previous program. The garbage collector is still my primary suspect, I > simply don't know how to explain the graph otherwise. I don't see a curve in the profile on my machine: http://www.haskell.org/~simonmar/main.exe.ps This was on a Windows box, I also get the same results on a Linux machine. I was only able to test with -M600m on the Windows box. Do you need more to get the curve? Cheers, Simon

2 1

-split-objs and --make together?
by Dimitry Golubovsky 29 Jul '05

29 Jul '05

Is it possible to use --make when one or more of modules involved are split? I tried this with ghc 6.2.2 I do the following (doesn't work) Suppose we have syscall.hs and UNISTD_H.hs: the first contains Main and imports UNISTD_H, and the second is to be split. mkdir UNISTD_H_split ghc -c -split-objs UNISTD_H.hs ar cq libUNISTD_H.a UNISTD_H_split/*.o I get the library and the .hi file: no .o file ghc --make syscall.hs libUNISTD_H.a -o syscall It tries to recompile UNISTD_H.hs because there is no UNISTD_H.o, and links against the object file, not the library When I try ghc syscall.hs libUNISTD.a -o syscall it works, and only syscall.hs is recompiled, but I cannot use --make. I tried to name the library UNISTD_H.a, same result. Is it possible to get ghc recognize the library same way as if it were an object file (which corresponds to the interface file) and use --make? PS1 This is for that experiment with hsffig when the header file contains functions undefined in the default glibc. Linking against the library with split objs works fine, and executable size is much smaller, and unresolved references are gone. Only ability to use --make is missing here. PS2 It would be nice if the perl script splitting the assembler code also created the directory where the split object files go. Dimitry Golubovsky Middletown, CT

1 0

Re: more on GHC 6.4 Debian packages
by Wolfgang Jeltsch 28 Jul '05

28 Jul '05

Am Donnerstag, 28. Juli 2005 20:46 schrieb Iavor Diatchki: > Hello, > > On 7/28/05, Wolfgang Jeltsch <wolfgang(a)jeltsch.net> wrote: > > Am Donnerstag, 28. Juli 2005 16:58 schrieb Ian Lynagh: > [...] > > > There are 6.4 packages in unstable. They aren't installable in > > > unstable, but I think they should be installable on a stable system. > > > > The unstable packages aren't installable on unstable but on stable? > > That's cool. :-) I will try to install them on stable. > > I am relatively new to the Debian distribution so appologies if my > question is silly (I don't understand the above statement). Debian has always the so-called distributions unstable, testing and stable. The package ghc6 from stable contains GHC 6.2.2 while the package ghc6 from unstable contains GHC 6.4. The package ghc6 from unstable isn't installable under Debian unstable but under Debian stable. > What do I need to do to use the "official" ghc 6.4 Debian packages? You may go to http://packages.debian.org/, download the ghc6 package from unstable manually (and possibly also ghc6-doc, ghc6-libsrc, etc.) and install it (them) via dpkg -i. I cannot tell you whether the unstable packages work under testing. > I am using the "testing" distribution, which appears to have something > called ghc-cvs and ghc 6.2 but not ghc 6.4. I didn't know about ghc-cvs so far. The idea of ghc-cvs looks a bit ugly. Why would the Debian project want to have an unstable GHC version in stable? In addition I wonder why ghc-cvs from stable is one year old. > What I did at the moment was to convert the ghc rpm into a debian package > using a tool called 'alien', and this worked without problems. Before doing this, I would always try to get packages specifically made for Debian. > I am confused as to why ghc 6.4 is considered less stable then ghc-cvs > which presumably is changing more or less all the time. That's related to the problem, I pointed out above. Even the ghc-cvs in stable should normally be more unstable than an official release of GHC in unstable. However, once it is in stable, ghc-cvs doesn't change its version anymore. That's the intention behind stable. (Well, Debian uses the term "stable" more to attest that version numbers don't change than to attest that the software runs stable, I think.) > -Iavor Best regards, Wolfgang

2 1

RE: switching off let-floating
by Simon Marlow 28 Jul '05

28 Jul '05

On 28 July 2005 10:41, Wolfgang Jeltsch wrote: > Am Mittwoch, 27. Juli 2005 21:39 schrieben Sie: >> "Wolfgang Jeltsch" <wolfgang(a)jeltsch.net> writes: >> >> .... >> >>> But how do I switch off let-floating? >> >> -fno-full-laziness, I believe. > > I just tried using this flag. Result: GHC 6.2.2 doesn't recognize > it. :-( That switch was added in 6.4. In 6.2 and earlier, there's no way to turn off full-laziness other than -Onot, as far as I'm aware. Cheers, Simon

4 7

GHC 6.4 Debian packages (was: Re: switching off let-floating)
by Wolfgang Jeltsch 28 Jul '05

28 Jul '05

Am Donnerstag, 28. Juli 2005 16:45 schrieben Sie: > On Thu, 28 Jul 2005, Wolfgang Jeltsch wrote: > (snip) > > > Related question: Can anybody tell me when there will be Debian packages > > of GHC 6.4 available? > > See http://packages.debian.org/unstable/devel/ghc6 > > -- Mark Okay, I wasn't precise enough. What I'm looking for are GHC 6.4 packages for Debian stable. Alas, the GHC website as well as Haskell Experimental or whatever it is called now don't have a respective backport. Regards, Wolfgang

1 0

switching off let-floating
by Wolfgang Jeltsch 28 Jul '05

28 Jul '05

Hello, I thought I had asked this before but I cannot find a corresponding mail from me in the archives and, in addition, cannot remember to have gotten an answer. The library documentation of unsafePerformIO in http://haskell.org/ghc/docs/latest/html/libraries/base/ System.IO.Unsafe.html#v%3AunsafePerformIO says: "Make sure that the[?] either you switch off let-floating, [...]". But how do I switch off let-floating? Regards, Wolfgang

2 4

RE: turning off garbage collection
by Simon Marlow 28 Jul '05

28 Jul '05

On 27 July 2005 05:48, Bernard Pope wrote: > I'm fairly sure this has been asked before (and recently too), but I > can't seem to find it in the archives. Sorry for the repetition. > > Is there a way I can turn garbage collection off in GHC? I'm happy to > hack the runtime and comment out code if that is the only way. I just > don't know which code. > > I want to see what effect GC is having on programs that run under > buddha. I suspect it is costing me a lot in time traversing large > amounts of data that the debugger is hanging on to. You can always see the proportion of GC time by using +RTS -sstderr. You can also set a gigantic allocation area with +RTS -A1G. Really disabling GC would entail just allocating more and more memory in the allocation area, and the RTS doesn't know how to do that at the moment (the alloc area is more or less fixed between GCs). Also, you'd very quickly eat all the swap in your system, so perhaps this isn't what you want. Cheers, Simon

2 1

turning off garbage collection
by Bernard Pope 27 Jul '05

27 Jul '05

Hi, I'm fairly sure this has been asked before (and recently too), but I can't seem to find it in the archives. Sorry for the repetition. Is there a way I can turn garbage collection off in GHC? I'm happy to hack the runtime and comment out code if that is the only way. I just don't know which code. I want to see what effect GC is having on programs that run under buddha. I suspect it is costing me a lot in time traversing large amounts of data that the debugger is hanging on to. Cheers, Bernie.

1 1

RE: ghc-pkg query feature request
by Simon Marlow 27 Jul '05

27 Jul '05

On 26 July 2005 09:23, Lennart Kolmodin wrote: > When working with multiple versions of a package you often want to use > the most recent version. > A feature that would help for package systems is to be able to query > ghc-pkg for the best version, or to list the versions in increasing > version order. > > Something like: > > $ ghc-pkg best_version Cabal > Cabal-1.1.1 > > $ ghc-pkg versions Cabal > Cabal-1.0 Cabal-1.1.1 > > You get the picture. Obviously, this feature isn't critical since we > could use 'ghc-pkg list' and work our way from there, but it would be > nice. If you like the idea we might send you a patch. > What do you think? Sure, I'll commit a patch if you send one. Perhaps $ ghc-pkg list Cabal Cabal-1.0 Cabal-1.1.1 would be the simplest extension? Cheers, Simon

2 1