Front page | perl.perl5.porters |
Postings from March 2003
On Fri, Mar 14, 2003 at 05:54:53PM +0100, Rafael Garcia-Suarez wrote:
> "Philippe 'BooK' Bruhat (via RT)" <perlbug-followup@perl.org> wrote:
> > The following short piece of code is broken.
> >
> > $ cat print.pl
> > #!/usr/bin/perl
> > my %data = ( foo => 'bar' );
> > print(
> > $data{foo});
...
> > $ perl5.8.0 -MO=Deparse print.pl
> > print.pl syntax OK
> > my(%data) = ('foo', 'bar');
> > print $main::data{'foo'};
>
> That "main::" isn't normal. %data is lexical.
>
> That's a parsing problem. While trying to disambiguate between the
> filehandle/indirect object "$data" and the value "$data{foo}" the
> parser gets confused and forgets to look in the pad. Sticking a
> STDOUT corrects the problem.
It does look in the pad. But it tries to snap '$data' not '%data'.
The '\n' causes yylex() to switch to the 0 case, and consequently
clobber PL_last_lop. So the test at toke.c:3473 will fail - and
'$' won't be changed into '%'.
Yet another skipspace() won't hurt :)
Regards
Adi
----------------------------------------------------------------
--- /arc/perl-current/toke.c 2003-03-09 17:03:05.000000000 +0200
+++ toke.c 2003-03-14 23:00:23.000000000 +0200
@@ -3088,6 +3088,7 @@ Perl_yylex(pTHX)
PL_oldbufptr = PL_oldoldbufptr; /* allow print(STDOUT 123) */
else
PL_expect = XTERM;
+ s = skipspace(s);
TOKEN('(');
case ';':
CLINE;
Thread Previous
|
Thread Next