runtime now looks at the SVf_UTF8 bit on the SV to decide

whether to use widechar semantics; lexer and RE engine continue
to need "use utf8" to enable unicode awareness in literals
and patterns (TODO: this needs to be fixed); $1 et al are marked
SvUTF8 if the pattern was compiled for utf8 (TODO: propagating
it from the data is probably better)

p4raw-id: //depot/perl@4930

1 files changed, 2 insertions, 0 deletions

diff --git a/op.c b/op.c
index 02edc8d9f1..fdfdf27b93 100644
--- a/op.c
+++ b/op.c
@@ -2867,6 +2867,8 @@ Perl_pmruntime(pTHX_ OP *o, OP *expr, OP *repl)
 	    p = SvPV(pat, plen);
 	    pm->op_pmflags |= PMf_SKIPWHITE;
 	}
+	else if (DO_UTF8(pat))
+	    pm->op_pmdynflags |= PMdf_UTF8;
 	pm->op_pmregexp = CALLREGCOMP(aTHX_ p, p + plen, pm);
 	if (strEQ("\\s+", pm->op_pmregexp->precomp))
 	    pm->op_pmflags |= PMf_WHITE;