Change 33529 by [EMAIL PROTECTED] on 2008/03/14 11:50:59
Integrate:
[ 33500]
Oops. Change 33499 forgot to add the new Makefile.PL to the MANIFEST.
[ 33504]
Correct two spelling mistakes, and reformat some comments.
Affected files ...
... //depot/maint-5.10/perl/MANIFEST#14 integrate
... //depot/maint-5.10/perl/perlio.c#3 integrate
Differences ...
==== //depot/maint-5.10/perl/MANIFEST#14 (text) ====
Index: perl/MANIFEST
--- perl/MANIFEST#13~33525~ 2008-03-14 03:11:45.000000000 -0700
+++ perl/MANIFEST 2008-03-14 04:50:59.000000000 -0700
@@ -898,6 +898,7 @@
ext/Module/Pluggable/lib/Devel/InnerPackage.pm Find inner packages
ext/Module/Pluggable/lib/Module/Pluggable/Object.pm Module::Pluggable
ext/Module/Pluggable/lib/Module/Pluggable.pm Module::Pluggable
+ext/Module/Pluggable/Makefile.PL Module::Pluggable
ext/NDBM_File/hints/cygwin.pl Hint for NDBM_File for named architecture
ext/NDBM_File/hints/dec_osf.pl Hint for NDBM_File for named architecture
ext/NDBM_File/hints/dynixptx.pl Hint for NDBM_File for named
architecture
==== //depot/maint-5.10/perl/perlio.c#3 (text) ====
Index: perl/perlio.c
--- perl/perlio.c#2~33501~ 2008-03-12 13:05:05.000000000 -0700
+++ perl/perlio.c 2008-03-14 04:50:59.000000000 -0700
@@ -3148,20 +3148,20 @@
if (!invalidate) {
#ifdef USE_ITHREADS
MUTEX_LOCK(&PL_perlio_mutex);
- /* Right. We need a mutex here because for a brief while we will
- have the situation that fd is actually closed. Hence if a
- second thread were to get into this block, its dup() would
- likely return our fd as its dupfd. (after all, it is closed).
+ /* Right. We need a mutex here because for a brief while we
+ will have the situation that fd is actually closed. Hence if
+ a second thread were to get into this block, its dup() would
+ likely return our fd as its dupfd. (after all, it is closed)
Then if we get to the dup2() first, we blat the fd back
(messing up its temporary as a side effect) only for it to
then close its dupfd (== our fd) in its close(dupfd) */
/* There is, of course, a race condition, that any other thread
trying to input/output/whatever on this fd will be stuffed
- for the duraction of this little manoeuver. Perhaps we should
- hold an IO mutex for the duration of every IO operation if
- we know that invalidate doesn't work on this platform, but
- that would suck, and could kill performance.
+ for the duration of this little manoeuvrer. Perhaps we
+ should hold an IO mutex for the duration of every IO
+ operation if we know that invalidate doesn't work on this
+ platform, but that would suck, and could kill performance.
Except that correctness trumps speed.
Advice from klortho #11912. */
End of Patch.
