Front page | perl.perl5.porters |
Postings from November 2008
On Fri, Nov 28, 2008 at 06:01:59PM +0000, Nicholas Clark wrote:
> As best I can tell, this is only with ithreads.
>
> Vanilla 5.8.8 (or with most recent Time::HiRes from CPAN), NYTProf 2.07, or
> with most recent from svn:
>
> $ ~/Sandpit/588ig/bin/perl5.8.8 -w -T -d:NYTProf -MTime::HiRes -e0
> No, I don't know why. Or, more importantly, what chain of events is causing
> PL_tainted to become set, and therefore what the cause is, and therefore
It's caused by this change:
http://public.activestate.com/cgi-bin/perlbrowse/p/28182
Specifically the integration of this:
Change 27176 by nicholas@nicholas-fangorn on 2006/02/13 21:46:13
Ensure that public I, N and P flags are off when SvTAINT is called on
something that already has taint magic.
Affected files ...
... //depot/perl/sv.c#1141 edit
Differences ...
==== //depot/perl/sv.c#1141 (text) ====
@@ -4242,8 +4242,13 @@
/* sv_magic() refuses to add a magic of the same 'how' as an
existing one
*/
- if (how == PERL_MAGIC_taint)
+ if (how == PERL_MAGIC_taint) {
mg->mg_len |= 1;
+ /* Any scalar which already had taint magic on which someone
+ (erroneously?) did SvIOK_on() or similar will now be
+ incorrectly sporting public "OK" flags. */
+ SvFLAGS(sv) &= ~(SVf_IOK|SVf_NOK|SVf_POK);
+ }
return;
}
}
Curiously I can't get it to fail in maint-5.10. I thought that it should do.
Nicholas Clark
Thread Previous
|
Thread Next