Change 34265 by [EMAIL PROTECTED] on 2008/09/04 14:27:46
Integrate:
[ 34085]
Subject: [PATCH] fix debop with pad const
From: Reini Urban <[EMAIL PROTECTED]>
Date: Tue, 24 Jun 2008 12:07:01 +0200
Message-ID: <[EMAIL PROTECTED]>
[ 34086]
Skip one test that was failing with DEBUGGING and threads,
because after change #34085 the IV value of the OP_CONST
isn't dumped anymore.
[ 34087]
Subject: [PATCH] Re: split smoke for threaded -DDEBUGGING builds
From: Reini Urban <[EMAIL PROTECTED]>
Date: Wed, 25 Jun 2008 17:30:08 +0200
Message-ID: <[EMAIL PROTECTED]>
Plus revert #34086
Affected files ...
... //depot/maint-5.10/perl/dump.c#9 integrate
... //depot/maint-5.10/perl/t/op/split.t#2 integrate
Differences ...
==== //depot/maint-5.10/perl/dump.c#9 (text) ====
Index: perl/dump.c
--- perl/dump.c#8~33946~ 2008-05-28 16:09:01.000000000 -0700
+++ perl/dump.c 2008-09-04 07:27:46.000000000 -0700
@@ -1946,7 +1946,14 @@
Perl_deb(aTHX_ "%s", OP_NAME(o));
switch (o->op_type) {
case OP_CONST:
- PerlIO_printf(Perl_debug_log, "(%s)", SvPEEK(cSVOPo_sv));
+ /* With ITHREADS, consts are stored in the pad, and the right pad
+ * may not be active here, so check.
+ * Looks like only during compiling the pads are illegal.
+ */
+#ifdef USE_ITHREADS
+ if ((((SVOP*)o)->op_sv) || !IN_PERL_COMPILETIME)
+#endif
+ PerlIO_printf(Perl_debug_log, "(%s)", SvPEEK(cSVOPo_sv));
break;
case OP_GVSV:
case OP_GV:
End of Patch.
