One more UTF-8 fix from Inaba Hiroto.

p4raw-id: //depot/perl@8415

diff --git a/pp.c b/pp.c
index 2d49cf1..784c7bf 100644 (file)
--- a/pp.c
+++ b/pp.c
@@ -5873,7 +5873,8 @@ PP(pp_split)
            s = m;
        }
     }
-    else if ((rx->reganch & RE_USE_INTUIT) && !rx->nparens
+    else if (do_utf8 == ((rx->reganch & ROPT_UTF8) != 0) &&
+            (rx->reganch & RE_USE_INTUIT) && !rx->nparens
             && (rx->reganch & ROPT_CHECK_ALL)
             && !(rx->reganch & ROPT_ANCH)) {
        int tail = (rx->reganch & RE_INTUIT_TAIL);

diff --git a/regexec.c b/regexec.c
index dca2a44..e85f45f 100644 (file)
--- a/regexec.c
+++ b/regexec.c
@@ -443,8 +443,6 @@ Perl_re_intuit_start(pTHX_ regexp *prog, SV *sv, char *strpos,
 #endif
 
   restart:
-    other_last = Nullch;
-
     /* Find a possible match in the region s..strend by looking for
        the "check" substring in the region corrected by start/end_shift. */
     if (flags & REXEC_SCREAM) {
@@ -1462,7 +1460,8 @@ Perl_regexec_flags(pTHX_ register regexp *prog, char *stringarg, register char *
            PL_reg_ganch = strbeg;
     }
 
-    if (!(flags & REXEC_CHECKED) && prog->check_substr != Nullsv) {
+    if (do_utf8 == (UTF!=0) &&
+       !(flags & REXEC_CHECKED) && prog->check_substr != Nullsv) {
        re_scream_pos_data d;
 
        d.scream_olds = &scream_olds;

diff --git a/t/pragma/utf8.t b/t/pragma/utf8.t
index 8e4d296..9137f36 100755 (executable)
--- a/t/pragma/utf8.t
+++ b/t/pragma/utf8.t
@@ -10,7 +10,7 @@ BEGIN {
     }
 }
 
-print "1..105\n";
+print "1..106\n";
 
 my $test = 1;
 
@@ -554,3 +554,13 @@ sub nok_bytes {
     print "ok $test\n";
     $test++;                                   # 105
 }
+
+{
+    use utf8;
+
+    my @a = split(/\xFE/, "\xFF\xFE\xFD");
+
+    print "not " unless @a == 2 && $a[0] eq "\xFF" && $a[1] eq "\xFD";
+    print "ok $test\n";
+    $test++;                                   # 106
+}