[gambit-list] Re: LAML for Gambit-C 4, gsi

Thomas Hafner hafner at sdf-eu.org
Fri Apr 1 12:54:43 EST 2005


Hello,

is anybody else interested in running Kurt Nørmark's LAML with Gambit
4? Kurt agrees that I forward some of our conversation about this.

I'll upload the attachment that I was talking about to
<http://hafner.sdf-eu.org/pool/laml-for-gambit4.tgz>

Thomas Hafner <thomas at hafner.nl.eu.org> wrote/schrieb <20050323194538.ECC4991B530 at schnucke.hafner.nl.eu.org>:

> Hello Mr. Nørmark,
> 
> Kurt Noermark <normark at cs.aau.dk> wrote/schrieb <yeg64zlz665.fsf at homer.cs.aau.dk>:
> 
> > Thomas Hafner <thomas at hafner.nl.eu.org> writes:
> > 
> > I will like to support your work on establishing a healthy LAML support
> > for Gambit-C. In the end, I would like to receive your setup files, and
> > include them in forthcoming LAML distributions.
> 
> Thanks for your offer to support me. I've attached an archive
> containing my first successfull configuration files for calling LAML
> from a command prompt.
> 
> I've used Gambit Version 4.0 beta 11 (the most recent released one
> would be beta 12).
> 
> Output of  uname -a is:
> Linux schnucke 2.4.29 #2 Mit Mär 9 19:35:04 CET 2005 i686 GNU/Linux
> 
> My OS is Debian 3.1.
> 
> > Have you looked at the Emacs stuff, more specifically how to activate
> > LAML via Emacs? Personally, I think this the most important part. 
> > An this part will certainly be part of a "full LAML support" for Gambit. 
> 
> Although I'm using Emacs, I'm not used to program in elisp. I noticed,
> that the procedure start-process is used to call LAML out of emacs.
> That makes me think about what to pass as arguments. You may wonder
> why this should be a problem. The promlem that I've encountered is
> that the Gambit interpreter (gsi or gsi-script) may treat command
> arguments as another files to be loaded. That depends on the first
> line of first file to be interpreted. I haven't found another means
> (e.g. command line option) to influence that behaviour than by writing
> that first line. That works like follows, using bash rather than a
> temporary file.
> 
> - If bash is called out of sh, just to demonstrate what have to be
> passed as arguments:
> 
> #!/bin/sh
> 
> exec bash -c 'gsi-script <(echo '\''#!/usr/bin/env gsi-script
> ;; begin of scheme code
> ;;...
> ;; end of scheme code
> '\'') "$@"' -- 'foo' 'bar' ...
> 
> - Speaking in terms of man 3 exec:
> 
> path[] shall be:
> bash
> 
> argv[0] shall be:
> -c
> 
> argv[1] shall be:
> gsi-script <(echo '#!/usr/bin/env gsi-script
> ;; begin of scheme code
> ;;...
> ;; end of scheme code
> ') "$@"
> 
> argv[2] shall be:
> --
> 
> argv[3] shall be:
> foo
> 
> argv[4] shall be:
> bar
> 
> a.s.o.
> 
> It's not very nice to depend from bash, because some systems (e.g.
> NetBSD) don't have it installed per default. But it's more elegant
> than a solution using a temporary file, I think. Maybe one day I'll be
> able to include that into laml-execute.el.
> 
> Regards
>   Thomas Hafner
> 
> -- 
> ___ http://hafner.sdf-eu.org/ ___
> ___ mailto:thomas@ ______________
> ____________ hafner.nl.eu.org ___

Kurt Noermark <normark at cs.aau.dk> wrote/schrieb <yegmzsmq4oi.fsf at homer.cs.aau.dk>:

> Thomas Hafner <thomas at hafner.nl.eu.org> writes:
> 
> Thomas,
> 
> I will put your compatibility file and your exec file into the LAML distribution,
> and I will notice that you have done the work.
> 
> With respect to starting LAML from Emacs: It is my experience that it can be 
> quite difficult to know the command line option well enough to start LAML successfully
> from Emacs. It took me some time to play with all the options when I handled
> MzScheme, Guile, and others. I do not know Gambit at all, and as such it will be very
> difficult for me (and most certainly time consuming) to set this up.
> 
> We will leave this aspect unimplemented for Gambit.
> 
> > Hello Mr. Nørmark,
> > 
> > Kurt Noermark <normark at cs.aau.dk> wrote/schrieb <yeg64zlz665.fsf at homer.cs.aau.dk>:
> > 
> > > Thomas Hafner <thomas at hafner.nl.eu.org> writes:
> > > 
> > > I will like to support your work on establishing a healthy LAML support
> > > for Gambit-C. In the end, I would like to receive your setup files, and
> > > include them in forthcoming LAML distributions.
> > 
> > Thanks for your offer to support me. I've attached an archive
> > containing my first successfull configuration files for calling LAML
> > from a command prompt.
> > 
> > I've used Gambit Version 4.0 beta 11 (the most recent released one
> > would be beta 12).
> > 
> > Output of  uname -a is:
> > Linux schnucke 2.4.29 #2 Mit Mär 9 19:35:04 CET 2005 i686 GNU/Linux
> > 
> > My OS is Debian 3.1.
> > 
> > > Have you looked at the Emacs stuff, more specifically how to activate
> > > LAML via Emacs? Personally, I think this the most important part. 
> > > An this part will certainly be part of a "full LAML support" for Gambit. 
> > 
> > Although I'm using Emacs, I'm not used to program in elisp. I noticed,
> > that the procedure start-process is used to call LAML out of emacs.
> > That makes me think about what to pass as arguments. You may wonder
> > why this should be a problem. The promlem that I've encountered is
> > that the Gambit interpreter (gsi or gsi-script) may treat command
> > arguments as another files to be loaded. That depends on the first
> > line of first file to be interpreted. I haven't found another means
> > (e.g. command line option) to influence that behaviour than by writing
> > that first line. That works like follows, using bash rather than a
> > temporary file.
> > 
> > - If bash is called out of sh, just to demonstrate what have to be
> > passed as arguments:
> > 
> > #!/bin/sh
> > 
> > exec bash -c 'gsi-script <(echo '\''#!/usr/bin/env gsi-script
> > ;; begin of scheme code
> > ;;...
> > ;; end of scheme code
> > '\'') "$@"' -- 'foo' 'bar' ...
> > 
> > - Speaking in terms of man 3 exec:
> > 
> > path[] shall be:
> > bash
> > 
> > argv[0] shall be:
> > -c
> > 
> > argv[1] shall be:
> > gsi-script <(echo '#!/usr/bin/env gsi-script
> > ;; begin of scheme code
> > ;;...
> > ;; end of scheme code
> > ') "$@"
> > 
> > argv[2] shall be:
> > --
> > 
> > argv[3] shall be:
> > foo
> > 
> > argv[4] shall be:
> > bar
> > 
> > a.s.o.
> > 
> > It's not very nice to depend from bash, because some systems (e.g.
> > NetBSD) don't have it installed per default. But it's more elegant
> > than a solution using a temporary file, I think. Maybe one day I'll be
> > able to include that into laml-execute.el.
> > 
> > Regards
> >   Thomas Hafner
> > 
> > -- 
> > ___ http://hafner.sdf-eu.org/ ___
> > ___ mailto:thomas@ ______________
> > ____________ hafner.nl.eu.org ___
> 
> -- 
> Kurt Nørmark                          Phone:        +45 9635 8080
> Department of Computer Science        Direct phone: +45 9635 8896
> Aalborg University                    Fax:          +45 9815 9889
> Fredrik Bajers Vej 7, Building E      E-mail: normark at cs.auc.dk
> DK-9220 Aalborg Ø                     WWW: http://www.cs.auc.dk/~normark/
> DENMARK                               Office: E4-208
> 

Regards
  Thomas

-- 
Alter macht immer weiß, aber nicht immer weise.
		-- Sprichwort



More information about the Gambit-list mailing list