[Gcl-devel] Re: Backquote bug in GCL? Help and/or comment requested.

[Camm Maguire]

Greetings, and thanks!

Masami Hagiya <address@hidden> writes:

> From: Robert Boyer <address@hidden>
> Subject: Re: Backquote bug in GCL? Help and/or comment requested.
> Date: Wed, 15 Mar 2006 08:19:15 -0600
> 
> > Dear Prof. Hagiya,
> > 
> > Thanks very much for your quick reply, which I am ccing below.  It turns out
> > the correct address for Guy Steele is now address@hidden
> > 
> > Thanks for the backquote fix.  It worked.
> 
> Thank you very much for checking it.
> 
> > A question.  By analogy, should one also delete the immediately following 
> > if,
> > in the NCONC case?  Probably a stupid question.
> 
> Yes.  You are right.
> 
> The old code almost always work.
> It transforms `(+ ,@x) into something like (cons '+ x).
> But for the example, it should transform `(+ ,@x) into (append '(+) x)
> in order to handle the internal ,@.
> 
> It would be nice if the old code is used if ,@ is not nested.
> 

How about this?

Index: backq.c
===================================================================
RCS file: /cvsroot/gcl/gcl/o/backq.c,v
retrieving revision 1.7
diff -u -r1.7 backq.c
--- backq.c     14 Oct 2005 22:41:24 -0000      1.7
+++ backq.c     15 Mar 2006 15:49:26 -0000
@@ -131,8 +131,11 @@
 
                case APPEND:
                        if (vs_head == Cnil) {
-                               vs_popp;
-                               return(EVAL);
+                         vs_popp;
+                         if (!consp(vs_head) ||
+                             vs_head->c.c_car!=siScomma_at)
+                           return(EVAL);
+                         vs_push(Cnil);
                        }
                        kwote_cdr();
                        make_list;
@@ -140,8 +143,11 @@
 
                case NCONC:
                        if (vs_head == Cnil) {
-                               vs_popp;
-                               return(EVAL);
+                         vs_popp;
+                         if (!consp(vs_head) ||
+                             vs_head->c.c_car!=siScomma_dot)
+                           return(EVAL);
+                         vs_push(Cnil);
                        }
                        kwote_cdr();
                        make_list;

Take care,



> -- Masami
> 
> > Thanks a lot,
> > 
> > Bob
> > 
> > -------------------------------------------------------------------------------
> > 
> > Date: Wed, 15 Mar 2006 23:06:45 +0900 (JST)
> > To: address@hidden
> > Cc: address@hidden, address@hidden, address@hidden,
> >         address@hidden, address@hidden,
> >         address@hidden
> > Subject: Re: Backquote bug in GCL? Help and/or comment requested.
> > From: Masami Hagiya <address@hidden>
> > 
> > As I look over the code, the problem seems to be in the following
> > fragment beginning with the 132nd line of backq.c.
> > 
> >             case APPEND:
> >                     if (vs_head == Cnil) {
> >                             vs_popp;
> >                             return(EVAL);
> >                     }
> >                     kwote_cdr();
> >                     make_list;
> >                     return(APPEND);
> > 
> > I guess removing the if statement will work...
> > 
> > Sorry, I haven't checked it yet.
> > 
> > -- Masami
> > 
> 
> 
> 

-- 
Camm Maguire                                            address@hidden
==========================================================================
"The earth is but one country, and mankind its citizens."  --  Baha'u'llah