February 2006 - Glasgow-haskell-users

Re: inside the GHC code generator
by Zdenek Dvorak 27 Feb '06

27 Feb '06

Hello, > > SPJ> tail calls is another, > > > > nowadays gcc optimize tail calls > > I found this thesis on the subject by Andreas Bauer: > > "Compilation of Functional Programming Languages using GCC -- Tail > Calls" > > Do you know if it is his work that was incorporated into GCC? no. Also, it is based on now obsolete version of the GCC (most of the tail call optimization is currently handled in a completely different way). If anyone has some concrete proposals on improvements of handling of tail calls in GCC that could help GHC, please use the gcc bugzilla (http://gcc.gnu.org/bugzilla/) to file an improvement request, and add me to CC, I will look what I can do. Zdenek

1 0

RE: inside the GHC code generator
by Simon Peyton-Jones 24 Feb '06

24 Feb '06

| The closest thing I've seen to a solution is the technique used in Urban | Boquist's thesis, which is to put a static table in the executable with | information about register and stack usage indexed by procedure return | point, and use that to unwind the stack and find roots. Every accurate garbage collector (including GHC's) uses a technique like this, but the solution is invariably tightly-coupled to the garbage collector, compiler and run-time system. A major feature of C-- is that it squarely addresses this problem in a *reusable* way. Simon

2 1

inside the GHC code generator
by Rene de Visser 24 Feb '06

24 Feb '06

The following link gives reasons for not generating via C http://groups.google.com/groups?hl=en&lr=&ie=UTF-8&selm=4zp8kn7xe.fsf_-_%40… Naturally a number of these are common lisp specific, however I think that Haskell and GCC are quite semantically different, so using GCC might prevent a lot of optimizations, that aren't possible in C, but would be in GHC. The OCAML-OPT backend is supposed to produce i386 assembly that is competitive with GCC. Maybe this could be ported to GHC? Rene.

5 16

RE: Re[3]: factorial: let's get ahead of jhc! :)
by Simon Peyton-Jones 24 Feb '06

24 Feb '06

| i have found my investiations in this area. that is the C-- code | generated for fac worker: | | Fac_zdwfac_entry() { | c1C0: | _s1BD = I32[Sp + 0]; | if (_s1BD != 1) goto c1C4; | R1 = I32[Sp + 4]; | Sp = Sp + 8; | jump (I32[Sp + 0]); | c1C4: | _s1BI = _s1BD * I32[Sp + 4]; | _s1BF = _s1BD - 1; | I32[Sp + 4] = _s1BI; | I32[Sp + 0] = _s1BF; | jump c1C0; | } Once we do the A/B split of the code generator that I referred to, we should get something like this from the A part fac(n,m) { if n!=1 goto L return(m) L: jump fac( n-1, n*m ) } The B part will introduce the Sp nonsense. I think you'll find this a much easier optimisation target. Simon -users

1 0

UTF-8 decoding error
by Christian Maeder 20 Feb '06

20 Feb '06

Hi, with ghc-6.5.20060201 I get a "UTF-8 decoding error" for latin1 characters in my string literals. Do I have to change my sources or can I set a certain environment variable? I have LANG=de_DE@euro and LC_CTYPE not set (which is ok for hugs) Cheers Christian

4 10

WinHugs for GHCi, prompt feature
by Neil Mitchell 13 Feb '06

13 Feb '06

Hi, I'm working on a Windows Haskell GUI, a lot like WinHugs (stealing a lot of the code from WinHugs), which I want to make cross-compiler happy - i.e. Hugs, GHCi and Yhc at least. [See http://www-users.cs.york.ac.uk/~ndm/projects/winhaskell.php for a really old screen shot] The way I am operating with Hugs is by spawning a Hugs process as a separate process, and redirecting the stdin, stdout and stderr to my application. The hardest bit is figuring out when Hugs has run a computation and finished, the way I am doing this currently is by setting the prompt to escape characters, then when my program sees these special escape characters it can determine that the computation has finished and let the user enter the next one. The problem with GHCi is that I haven't been able to set the prompt, and therefore I can't detect when the computation is over, which makes it quite hard to operate with. So what I would really like is any one of: 1) Ability to set the prompt in GHCi 2) Some other mechanism to detect when the computation has finished 3) A way under Windows to detect when a console program is waiting on user input (which I have been entirely unable to find, and may just not exist) WinHaskell is currently perfectly useable with Hugs, and maybe a month or so away from release. I intend to start a darcs repo on tomorrow, if anyone is interested in it. Thanks Neil

4 7

ld.exe distributed with Windows ghc-6.4.1 buggy?
by Bayley, Alistair 13 Feb '06

13 Feb '06

(MS Windows) GHC users, I've experienced a small problem with the version of ld that ships with GHC-6.4.1 for Windows. I used it to build a Postgres sample C program, and the resulting executable segfaults on a certain line. The same program works correctly when linked with the version of ld (a slightly older one) that comes with my Mingw installation. The respective ld versions are: Ghc-6.4.1: 2.15.91 20040904 Mingw: 2.13.90 20030111 For now I've just replaced the ld.exe in C:\ghc\ghc-6.4.1\gcc-lib\ with the Mingw one, and it's all good. Simons (when you return from holiday): is it worth persuing a narrower test case? (one that doesn't require you to install Postgres to reproduce :-) Alistair ***************************************************************** Confidentiality Note: The information contained in this message, and any attachments, may contain confidential and/or privileged material. It is intended solely for the person(s) or entity to which it is addressed. Any review, retransmission, dissemination, or taking of any action in reliance upon this information by persons or entities other than the intended recipient(s) is prohibited. If you received this in error, please contact the sender and delete the material from any computer. *****************************************************************

1 0

RE: newtype and rules
by Simon Peyton-Jones 12 Feb '06

12 Feb '06

| I am curious if rules pragmas are applied before or after newtype | desugaring? The matching function respects newtypes. | basically what I want to know is if I have a newtype of an | Int say, and make a rule acting on the newtype, will it be applied to | any Int as well? No, it shouldn't be. Simon

2 1

forall a (Ord a => a-> a) -> Int is an illegal type???
by Brian Hulley 12 Feb '06

12 Feb '06

Hi - I've been puzzling over section 7.4.9.3 of the ghc users manual for the past few months (!) and still can't understand why the following is an illegal type: forall a. ((Ord a => a-> a) -> Int) whereas (forall a. Ord a => a->a) -> Int is legal. I can understand why the second one *is* legal but I can't seem to understand why the first syntax is not just exactly the same thing even though the parse tree is different. Can anyone shed some light on this? Thanks, Brian.

7 14

GHC HQ shut down for a week
by Simon Peyton-Jones 10 Feb '06

10 Feb '06

Cherished users Simon and I are both on holiday next week, so you won't hear from GHC HQ during that time. Happy Haskell hacking! Simon

1 0