PATCH: pws-25: another line number problem

PATCH: pws-25: another line number problem

[Peter Stephenson]

Consider:

% fpath=(.)
% cat alf
[[ foo bar ]] && print That did\'t work

#
print foo
% autoload alf
% alf
alf: parse error: condition expected: foo [5]

The line number reported is the last line of the file.  The answer,
thankfully, is simple:  don't flush the input queue until the error message
has been printed.  In fact, we ought to be able to tell that the input
queue is a string and will be flushed automatically when it's popped off
the input stack.  Maybe I'll put this on my to-do list.  Maybe.

--- Src/parse.c.uln	Thu Jun 17 15:28:47 1999
+++ Src/parse.c	Wed Jul  7 10:49:02 1999
@@ -72,7 +72,7 @@
 
 #define YYERROR  { tok = LEXERR; return NULL; }
 #define YYERRORV { tok = LEXERR; return; }
-#define COND_ERROR(X,Y) do{herrflush();zerr(X,Y,0);YYERROR}while(0)
+#define COND_ERROR(X,Y) do{zerr(X,Y,0);herrflush();YYERROR}while(0)
 
 #define make_list()     allocnode(N_LIST)
 #define make_sublist()  allocnode(N_SUBLIST)
@@ -143,8 +143,8 @@
 	    yyerror();
 	    return NULL;
 	}
-	herrflush();
 	yyerror();
+	herrflush();
 	return NULL;
     } else {
 	l->right = par_event();


-- 
Peter Stephenson <pws@ibmth.df.unipi.it>       Tel: +39 050 844536
WWW:  http://www.ifh.de/~pws/
Dipartimento di Fisica, Via Buonarroti 2, 56127 Pisa, Italy

Re: PATCH: pws-25: another line number problem

[Peter Stephenson]

Peter Stephenson wrote:
> Consider:
> 
> % fpath=(.)
> % cat alf
> [[ foo bar ]] && print That did\'t work
> 
> #
> print foo
> % autoload alf
> % alf
> alf: parse error: condition expected: foo [5]
> 
> The line number reported is the last line of the file.  The answer,
> thankfully, is simple:  don't flush the input queue until the error message
> has been printed.

Gag.  Now I've found out why the error had to come second: if errflag is
set, history flushing won't work, so if the parse error occurred
interactively, you don't get the rest of the line in the history.  Let's
just try something like my other proposal:  don't do a full history flush
if reading from a string.  This *replaces* the previous patch.

--- Src/hist.c.uln2	Thu Jun 17 14:22:18 1999
+++ Src/hist.c	Wed Jul  7 11:39:09 1999
@@ -291,7 +291,7 @@
 void
 herrflush(void)
 {
-    while (!lexstop && inbufct)
+    while (!lexstop && inbufct && !strin)
 	hwaddc(ingetc());
 }
 

-- 
Peter Stephenson <pws@ibmth.df.unipi.it>       Tel: +39 050 844536
WWW:  http://www.ifh.de/~pws/
Dipartimento di Fisica, Via Buonarroti 2, 56127 Pisa, Italy