Don’t inline sub(){0; $x} if $x changes elsewhere

Some op trees will turn a sub into a constant even if they
are more than just a simple constant or lexical variable.  In
particular, a statement optimised away followed by a lexi-
cal variable is eligible for inlining.  As discussed in
<20141101170924.17763.qmail@lists-nntp.develooper.com>, make these
more complex op trees follow closure rules properly.  If the outer
lexical is used in lvalue context in places other than its declara-
tion, then we should forego inlining.

1 files changed, 5 insertions, 3 deletions

diff --git a/op.c b/op.c
index 68793c1aa4..b5a5b6df9d 100644
--- a/op.c
+++ b/op.c
@@ -7756,16 +7756,18 @@ Perl_op_const_sv(pTHX_ const OP *o, CV *cv, CV *outcv)
 				 nextstate
 				 padsv
 			 */
-			if (ckWARN_d(WARN_DEPRECATED)
-			 && OP_SIBLING(
+			if (OP_SIBLING(
 			     cUNOPx(cUNOPx(CvROOT(cv))->op_first)->op_first
 			    ) == o
 			 && !OP_SIBLING(o))
-			    Perl_warner(aTHX_ packWARN(WARN_DEPRECATED),
+			    Perl_ck_warner_d(aTHX_
+					      packWARN(WARN_DEPRECATED),
 					     "Constants from lexical "
 					     "variables potentially "
 					     "modified elsewhere are "
 					     "deprecated");
+			else
+			    return NULL;
 		    }
 		}
 		sv = PAD_BASE_SV(CvPADLIST(cv), o->op_targ);