I had something of a problem with Guile, because when it arrayifies a list
SCM it stringifies all elements of the list to convert them to SVs; this
caused me horrible problems since one of the elements of my list was a
continuation. Plus, it tried to convert the continuation to a double because
the last compare in stringify was wrong (scm_ functions return an SCM, not an
int return, plus it was testing for exact when it meant inexact). I've fixed
the compare (I think) and I've tweaked the code so that it converts the list
elements to Guile::SCM objects instead (since they can stringify later if
needed anyway). There may be a better way to do this, but the following patch
is what I'm using with some success -
*** guile_wrap.c.orig Thu Jan 1 22:05:34 2004
--- guile_wrap.c Thu Jan 1 22:05:36 2004
***************
*** 170,177 ****
// might make sense if Perl had car and cdr!
if (gh_list_p(scm)) {
AV *av = newAV();
do {
! av_push(av, newSVscm(SCM_CAR(scm)));
scm = SCM_CDR(scm);
} while(scm != SCM_EOL);
return newRV_noinc((SV*)av);
--- 170,183 ----
// might make sense if Perl had car and cdr!
if (gh_list_p(scm)) {
AV *av = newAV();
+ SV* sv;
+ SCM tmp;
do {
! sv = newSV(0);
! tmp = SCM_CAR(scm);
! scm_gc_protect_object(tmp);
! sv_setref_pv(sv, "Guile::SCM", (void*)tmp);
! av_push(av, sv);
scm = SCM_CDR(scm);
} while(scm != SCM_EOL);
return newRV_noinc((SV*)av);
***************
*** 181,189 ****
if (SCM_CONSP(scm)) {
// create a two-element array with the CAR and CDR of the pair
AV *av = newAV();
av_extend(av, 1);
! av_store(av, 0, newSVscm(SCM_CAR(scm)));
! av_store(av, 1, newSVscm(SCM_CDR(scm)));
return newRV_noinc((SV*)av);
}
--- 187,205 ----
if (SCM_CONSP(scm)) {
// create a two-element array with the CAR and CDR of the pair
AV *av = newAV();
+ SV* sv;
+ SCM tmp;
av_extend(av, 1);
! sv = newSV(0);
! tmp = SCM_CAR(scm);
! scm_gc_protect_object(tmp);
! sv_setref_pv(sv, "Guile::SCM", (void*)tmp);
! av_store(av, 0, sv);
! sv = newSV(0);
! tmp = SCM_CDR(scm);
! scm_gc_protect_object(tmp);
! sv_setref_pv(sv, "Guile::SCM", (void*)tmp);
! av_store(av, 1, sv);
return newRV_noinc((SV*)av);
}
***************
*** 192,198 ****
return newSVpvn(SCM_STRING_CHARS(scm),SCM_STRING_LENGTH(scm));
// floats
! if (scm_exact_p(scm))
return newSVnv(gh_scm2double(scm));
croak("Guile::newSVscm : Unknown non-immediate SCM type.");
--- 208,214 ----
return newSVpvn(SCM_STRING_CHARS(scm),SCM_STRING_LENGTH(scm));
// floats
! if (scm_inexact_p(scm) == SCM_BOOL_T)
return newSVnv(gh_scm2double(scm));
croak("Guile::newSVscm : Unknown non-immediate SCM type.");
======== END OF PATCH ==========
The reason I ran in to this is because I'm trying to get Guile working as a
control language for POE programs by using a function that throws an exception
with a continuation as one of the parameters, so the Guile execution pauses
but returns the continuation. I then interpret the other arguments of the
continuation as the details of a POE event to post, and when the results of
the event's processing are posted back to the Guile POE::Session it re-invokes
the continuation with them - so that the POE work is basically transparent
from within the Guile code. I've got a working REPL using POE::Wheel::ReadLine
(and a wrapper object which is started at the request of the initial Guile
thread) and am going to add some more wrappers so you can actually do
something with it. I'll lob a post out here when I've got code that does
stuff; meantime, if you want a look at what I've got so far feel free to
annoy me by e-mail :)
--
Bring me my etherkiller; Oh clouds unfold! / Bring me the magic smoke of desire
I shall not cease from mental fight / Nor shall my LART rest in my hand
Till we have buried the bodies / Of all the lusers in all this land
-- rpg, ASR [ My homepage is http://www.trout.me.uk/ ]
