This is the mail archive of the guile@sources.redhat.com mailing list for the Guile project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]

Re: Guile IMHO and a proposal.


> Mailing-List: contact guile-help@sourceware.cygnus.com; run by ezmlm
> Precedence: bulk
> List-Unsubscribe: <mailto:guile-unsubscribe-mvanier=bbb.caltech.edu@sources.redhat.com>
> List-Subscribe: <mailto:guile-subscribe@sources.redhat.com>
> List-Archive: <http://sources.redhat.com/ml/guile/>
> List-Post: <mailto:guile@sources.redhat.com>
> List-Help: <mailto:guile-help@sources.redhat.com>, <http://sources.redhat.com/ml/#faqs>
> Sender: guile-owner@sources.redhat.com
> Cc: rms@gnu.org, guile@sourceware.cygnus.com
> Organization: MCCME
> From: "Khimenko Victor" <guile@khim.sch57.msk.ru>
> Date: Mon, 10 Jul 2000 22:50:50 +0400 (MSD)
> Content-Type: text/plain; charset=us-ascii
> 
> 10-Jul-00 05:06 you wrote:
> >> *Guile is an embeddable version of a subset some implementation of Scheme.
> >> -As there is little documentation this makes for tough work anyone wanting
> >> to use Guile even if they know another implementation of Scheme.
> 
> > Actually, this is the one thing I found rather easy about guile.  I
> > haven't played with it in a while, but the documentation for creating
> > scheme procedures in c, and calling scheme procedures from c was quite
> > good, and it was very simple as well.  By reading the tutorial, I had
> > compiled a program that called scheme functions.
> 
> > The main drawback I see in Guile is the main() function thing.  I was
> > looking into writing a mod_scheme for apache using guile, and decided that
> > most people wouldn't use a source-hacked apache just to host an
> > expiremental model.  Although I know from this list that it is
> > theoretically possible to do GC without hacking main(), might it be
> > simpler to go to a reference-counting model?
> 
> If you need 100% portable solution - yes. If you want "only" support for
> Linux, *BSD, Win32 (and few proprietary *nix'es like Solaris and HP-UX)
> then you can create version of guile without main() hack in few weekends
> (on one weekend if you are lucky) - it'll be REALLY faster then adoption of
> reference-counting GC.
> 
> > I know refcounting doesn't take care of circular data structures, but is
> > that really a problem?
> 
> With scheme - yes.
> 
> > It hasn't hurt Perl any.  Anyway, this is probably out of the question - it
> > would probably change too many things.  However, if there were a volunteer
> > to move scheme from GC to refcounting, would there be any objections?
> 

Actually, perl also does mark-and-sweep gc at certain points (this is
relatively new and was introduced IIRC to deal with closures).  Since other
scripting languages (e.g. python) are moving more and more towards true GC,
this doesn't seem to be a good idea to me (not to mention that it violates
the definition of scheme at a very fundamental level).

Mike


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]