Bug in format/write #6624
Comments
From schopper@usm.uni-muenchen.deHi there, I belive, that we've found a bug in perl. We discovered it when changing from We found, that write no longer use the actual values of the variables A sample program would be: 1 #!/usr/lib/perl5/neu/usr-bin/perl -w This gives the wrong output: But when we comment out line 5 we get the right result: The old (correct) version of perl is:Summary of my perl5 (5.0 patchlevel 4 subversion 4) configuration: Characteristics of this binary (from libperl): The new (buggy) Version is:Summary of my perl5 (revision 5.0 version 8 subversion 0) configuration: Characteristics of this binary (from libperl): Thanks in advance for any help you can give. With best regards Rüdiger Schopper GPG/PGP-Key: _/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/ |
From @iabynOn Mon, Jul 14, 2003 at 05:37:42AM -0000, Ruediger Schopper wrote:
The bug can be reduced to the simpler case: sub f ($); # Comment out to get right result! It's due to there being multiple CVs associated with f due to to the Since I'm currently working on trying to get sub declarations -- |
From schopper@usm.uni-muenchen.deHi Dave, thanks for the extremely fast response. I would like to know if we might be Our problem is, that we have to upgrade perl eventually, so we are in principle I don't want to annoy you, so just give us a short notice, if you like. Best regards Rüdiger
GPG/PGP-Key: _/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/ |
From @iabynOn Mon, Jul 21, 2003 at 05:43:39PM +0200, Ruediger Schopper wrote:
Thanks for the kind offer, but I don't think that giving money to me However, if you want to contribute financially to the development of Perl I am hoping to have a patch ready in 2-3 weeks time, but it's Of course, I may find that I can't fix it after all... Regards, Dave. -- |
From schopper@usm.uni-muenchen.deHi Dave, do you have any news concerning the subroutine/format bug I sent you a With best regards Rüdiger
GPG/PGP-Key: _/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/ |
From @iabynOn Tue, Sep 02, 2003 at 01:28:58PM +0200, Ruediger Schopper wrote:
I'm working on a opath, bug have got side-tracked with things more
-- |
From @iabynOn Tue, Sep 02, 2003 at 09:01:05PM +0100, Dave Mitchell wrote: Would have helped if I'd proof-read it before hitting the send key. I'm working on a patch, but have got side-tracked with things more -- |
From Alexander.stohr@gmx.deIt seems to me that this bug is still a pending one. #!/usr/bin/perl -w Results are: ARRAY(0x81317a4) 101 (loop) ARRAY(0x8131888) 102 (loop) |
From p5p@spam.wizbit.beOn Mon Jul 14 15:15:03 2003, davem wrote:
It appears that things got worse for 5.10.0/blead... perl-5.8.8 rt-22977.pl (Instead of 1 2) perl-5.9.0 rt-22977.pl perl-5.10.0 rt-22977.pl perl-blead rt-22977.pl |
From @andk
> It appears that things got worse for 5.10.0/blead... Binary search reveals... ----Program---- ----Output of .../pMZAcVt/perl-5.8.0@19648/bin/perl---- ----EOF ($?='0')---- ----EOF ($?='11')---- Change 19649 by rgs@rgs-home on 2003/05/31 18:33:07 Subject: [PATCH] jumbo closure patch broke formats -- |
From @nwc10Dave notes: Was a format/closure/newATTSUB bug in 5.8.8; in 5.10.0, maint, blead, it |
From chromatic@wgz.orgOn Friday 27 June 2008 12:04:09 Bram via RT wrote:
That's the same bug as RT #50528. -- c |
From zefram@fysh.orgAttached patch fixes bug #22977. make regen_perly after applying. -zefram |
From zefram@fysh.orgInline Patchdiff --git a/perly.y b/perly.y
index 18e5875..a61a6b3 100644
--- a/perly.y
+++ b/perly.y
@@ -511,7 +511,9 @@ peg : PEG
;
format : FORMAT startformsub formname block
- { SvREFCNT_inc_simple_void(PL_compcv);
+ {
+ CV *fmtcv = PL_compcv;
+ SvREFCNT_inc_simple_void(PL_compcv);
#ifdef MAD
$$ = newFORM($2, $3, $4);
prepend_madprops($1->tk_mad, $$, 'F');
@@ -521,6 +523,10 @@ format : FORMAT startformsub formname block
newFORM($2, $3, $4);
$$ = (OP*)NULL;
#endif
+ if (CvOUTSIDE(fmtcv) && !CvUNIQUE(CvOUTSIDE(fmtcv))) {
+ SvREFCNT_inc_simple_void(fmtcv);
+ pad_add_anon((SV*)fmtcv, OP_NULL);
+ }
}
;
diff --git a/t/comp/form_scope.t b/t/comp/form_scope.t
new file mode 100644
index 0000000..3ef891e
--- /dev/null
+++ b/t/comp/form_scope.t
@@ -0,0 +1,18 @@
+#!./perl
+#
+# Tests bug #22977. Test case from Dave Mitchell.
+
+print "1..2\n";
+
+sub f ($);
+sub f ($) {
+my $test = $_[0];
+write;
+format STDOUT =
+ok @<<<<<<<
+$test
+.
+}
+
+f(1);
+f(2); |
From @rgarcia2009/12/15 Zefram <zefram@fysh.org>:
Thanks, applied, regenerated, and MANIFEST completed. |
|
@rgs - Status changed from 'open' to 'resolved' |
From @cpansproutOn Tue Dec 15 02:49:25 2009, rgs@consttype.org wrote:
The commit id was 421f30e. The commit message did not explain: It follows Dave Mitchell’s -- Father Chrysostomos |
Migrated from rt.perl.org#22977 (status was 'resolved')
Searchable as RT22977$
The text was updated successfully, but these errors were encountered: