integrate cfgperl contents

p4raw-id: //depot/perl@3656

5 files changed, 95 insertions, 39 deletions

diff --git a/t/op/pat.t b/t/op/pat.t
index a086c12eaf..6312c75cea 100755
--- a/t/op/pat.t
+++ b/t/op/pat.t
@@ -282,14 +282,7 @@ eval qq("${context}y" =~ /(?<=$context)y/);
 print "not " if $@ !~ m%^\Q/(?<=\Ex+/: lookbehind longer than 255 not%;
 print "ok 71\n";
 
-# This one will fail when POSIX character classes do get implemented
-{
-	my $w;
-	local $^W = 1;
-	local $SIG{__WARN__} = sub{$w = shift};
-	eval q('a' =~ /[[:alpha:]]/);
-	print "not " if $w !~ /^\QCharacter class syntax [: :] is reserved/;
-}
+# removed test
 print "ok 72\n";
 
 # Long Monsters
diff --git a/t/op/re_tests b/t/op/re_tests
index 466fc856c9..cbcb7251b1 100644
--- a/t/op/re_tests
+++ b/t/op/re_tests
@@ -474,9 +474,37 @@ $(?<=^(a))	a	y	$1	a
 ([[=]+)	a=[b]=	y	$1	=[
 ([[.]+)	a.[b].	y	$1	.[
 [a[:xyz:	-	c	-	/[a[:xyz:/: unmatched [] in regexp
-[a[:xyz:]	-	c	-	/[a[:xyz:]/: unmatched [] in regexp
+[a[:xyz:]	-	c	-	Character class [:xyz:] unknown
 [a[:]b[:c]	abc	y	$&	abc
-([a[:xyz:]b]+)	pbaq	y	$1	ba
+([a[:xyz:]b]+)	pbaq	c	-	Character class [:xyz:] unknown
+[a[:]b[:c]	abc	y	$&	abc
+([[:alpha:]]+)	ABcd01Xy__--  ${nulnul}${ffff}	y	$1	ABcd
+([[:alnum:]]+)	ABcd01Xy__--  ${nulnul}${ffff}	y	$1	ABcd01Xy
+([[:ascii:]]+)	ABcd01Xy__--  ${nulnul}${ffff}	y	$1	ABcd01Xy__--  ${nulnul}
+([[:cntrl:]]+)	ABcd01Xy__--  ${nulnul}${ffff}	y	$1	${nulnul}
+([[:digit:]]+)	ABcd01Xy__--  ${nulnul}${ffff}	y	$1	01
+([[:graph:]]+)	ABcd01Xy__--  ${nulnul}${ffff}	y	$1	ABcd01Xy__--
+([[:lower:]]+)	ABcd01Xy__--  ${nulnul}${ffff}	y	$1	cd
+([[:print:]]+)	ABcd01Xy__--  ${nulnul}${ffff}	y	$1	ABcd01Xy__--  
+([[:punct:]]+)	ABcd01Xy__--  ${nulnul}${ffff}	y	$1	__--
+([[:space:]]+)	ABcd01Xy__--  ${nulnul}${ffff}	y	$1	  
+([[:word:]]+)	ABcd01Xy__--  ${nulnul}${ffff}	y	$1	ABcd01Xy__
+([[:upper:]]+)	ABcd01Xy__--  ${nulnul}${ffff}	y	$1	AB
+([[:xdigit:]]+)	ABcd01Xy__--  ${nulnul}${ffff}	y	$1	ABcd01
+([[:^alpha:]]+)	ABcd01Xy__--  ${nulnul}${ffff}	y	$1	01
+([[:^alnum:]]+)	ABcd01Xy__--  ${nulnul}${ffff}	y	$1	__--  ${nulnul}${ffff}
+([[:^ascii:]]+)	ABcd01Xy__--  ${nulnul}${ffff}	y	$1	${ffff}
+([[:^cntrl:]]+)	ABcd01Xy__--  ${nulnul}${ffff}	y	$1	ABcd01Xy__--  
+([[:^digit:]]+)	ABcd01Xy__--  ${nulnul}${ffff}	y	$1	ABcd
+([[:^lower:]]+)	ABcd01Xy__--  ${nulnul}${ffff}	y	$1	AB
+([[:^print:]]+)	ABcd01Xy__--  ${nulnul}${ffff}	y	$1	${nulnul}${ffff}
+([[:^punct:]]+)	ABcd01Xy__--  ${nulnul}${ffff}	y	$1	ABcd01Xy
+([[:^space:]]+)	ABcd01Xy__--  ${nulnul}${ffff}	y	$1	ABcd01Xy__--
+([[:^word:]]+)	ABcd01Xy__--  ${nulnul}${ffff}	y	$1	--  ${nulnul}${ffff}
+([[:^upper:]]+)	ABcd01Xy__--  ${nulnul}${ffff}	y	$1	cd01
+([[:^xdigit:]]+)	ABcd01Xy__--  ${nulnul}${ffff}	y	$1	Xy__--  ${nulnul}${ffff}
+[[:foo:]]	-	c	-	Character class [:foo:] unknown
+[[:^foo:]]	-	c	-	Character class [:^foo:] unknown
 ((?>a+)b)	aaab	y	$1	aaab
 (?>(a+))b	aaab	y	$1	aaa
 ((?>[^()]+)|\([^()]*\))+	((abc(ade)ufh()()x	y	$&	abc(ade)ufh()()x
diff --git a/t/op/regexp.t b/t/op/regexp.t
index 66b2d1c116..4ffe1362c6 100755
--- a/t/op/regexp.t
+++ b/t/op/regexp.t
@@ -47,6 +47,8 @@ seek(TESTS,0,0);
 $. = 0;
 
 $bang = sprintf "\\%03o", ord "!"; # \41 would not be portable.
+$ffff  = chr(0xff) x 2;
+$nulnul = "\0" x 2;
 
 $| = 1;
 print "1..$numtests\n# $iters iterations\n";
@@ -59,12 +61,16 @@ while (<TESTS>) {
     infty_subst(\$pat);
     infty_subst(\$expect);
     $pat = "'$pat'" unless $pat =~ /^[:']/;
-    $pat =~ s/\\n/\n/g;
     $pat =~ s/(\$\{\w+\})/$1/eeg;
+    $pat =~ s/\\n/\n/g;
+    $subject =~ s/(\$\{\w+\})/$1/eeg;
     $subject =~ s/\\n/\n/g;
+    $expect =~ s/(\$\{\w+\})/$1/eeg;
     $expect =~ s/\\n/\n/g;
     $expect = $repl = '-' if $skip_amp and $input =~ /\$[&\`\']/;
     $skip = ($skip_amp ? ($result =~ s/B//i) : ($result =~ s/B//));
+    # Certain tests don't work with utf8 (the re_test should be in UTF8)
+    $skip = 1 if ($^H &= ~0x00000008) && $pat =~ /\[:\^(alnum|print|word):\]/;
     $result =~ s/B//i unless $skip;
     for $study ('', 'study \$subject') {
  	$c = $iters;
@@ -75,7 +81,7 @@ while (<TESTS>) {
 	    last;  # no need to study a syntax error
 	}
 	elsif ( $skip ) {
-	    print "ok $. # Skipped: not fixed yet\n"; next TEST;
+	    print "ok $. # skipped\n"; next TEST;
 	}
 	elsif ($@) {
 	    print "not ok $. $input => error `$err'\n"; next TEST;
diff --git a/t/pragma/utf8.t b/t/pragma/utf8.t
index e5b3bb5cd3..5e467ae053 100755
--- a/t/pragma/utf8.t
+++ b/t/pragma/utf8.t
@@ -6,7 +6,7 @@ BEGIN {
     $ENV{PERL5LIB} = '../lib';
 }
 
-print "1..3\n";
+print "1..9\n";
 
 my $test = 1;
 
@@ -34,4 +34,37 @@ sub ok {
     s/([$rx])/"&#".ord($1).";"/eg; 
     ok $_, '>&#9786;<';
     $test++;
+
+    $_ = "alpha,numeric"; 
+    m/([[:alpha:]]+)/; 
+    ok $1, 'alpha';
+    $test++;
+
+    $_ = "alphaNUMERICstring";
+    m/([[:^lower:]]+)/; 
+    ok $1, 'NUMERIC';
+    $test++;
+
+    $_ = "alphaNUMERICstring";
+    m/(\p{Ll}+)/; 
+    ok $1, 'alpha';
+    $test++;
+
+    $_ = "alphaNUMERICstring"; 
+    m/(\p{Lu}+)/; 
+    ok $1, 'NUMERIC';
+    $test++;
+
+    $_ = "alpha,numeric"; 
+    m/([\p{IsAlpha}]+)/; 
+    ok $1, 'alpha';
+    $test++;
+
+    $_ = "alphaNUMERICstring";
+    m/([^\p{IsLower}]+)/; 
+    ok $1, 'NUMERIC';
+    $test++;
+
 }
+
+
diff --git a/t/pragma/warn/regcomp b/t/pragma/warn/regcomp
index 4b2f7ff2ba..6545778a38 100644
--- a/t/pragma/warn/regcomp
+++ b/t/pragma/warn/regcomp
@@ -8,9 +8,6 @@
 
 	/(?=a)?/
 
-  Character class syntax [: :] is reserved for future extensions
-	/[a[:xyz:]b]/
-
   Character class syntax [. .] is reserved for future extensions
   Character class syntax [= =] is reserved for future extensions
 
@@ -36,30 +33,29 @@ Strange *+?{} on zero-length expression at - line 4.
 # regcomp.c
 use warning 'unsafe' ;
 $_ = "" ;
-/[a[:xyz:]b]/;
-/[a[.xyz.]b]/;
-/[a[=xyz=]b]/;
+/[:alpha:]/;
+/[.bar.]/;
+/[=zog=]/;
+/[[:alpha:]]/;
+/[[.foo.]]/;
+/[[=bar=]]/;
+/[:zog:]/;
 no warning 'unsafe' ;
-/[a[:xyz:]b]/;
-/[a[.xyz.]b]/;
-/[a[=xyz=]b]/;
+/[:alpha:]/;
+/[.foo.]/;
+/[=bar=]/;
+/[[:alpha:]]/;
+/[[.foo.]]/;
+/[[=bar=]]/;
+/[:zog:]/;
+/[[:zog:]]/;
 EXPECT
-Character class syntax [: :] is reserved for future extensions at - line 4.
+Character class syntax [: :] belongs inside character classes at - line 4.
+Character class syntax [. .] belongs inside character classes at - line 5.
 Character class syntax [. .] is reserved for future extensions at - line 5.
+Character class syntax [= =] belongs inside character classes at - line 6.
 Character class syntax [= =] is reserved for future extensions at - line 6.
-########
-# regcomp.c
-use warning 'unsafe' ; 
-# use utf8 ; # Note this line should be uncommented when utf8 gets fixed.
-$_ = "" ;
-/[a[:xyz:]b]/;
-/[a[.xyz.]b]/;
-/[a[=xyz=]b]/;
-no warning 'unsafe' ;
-/[a[:xyz:]b]/;
-/[a[.xyz.]b]/;
-/[a[=xyz=]b]/;
-EXPECT
-Character class syntax [: :] is reserved for future extensions at - line 5.
-Character class syntax [. .] is reserved for future extensions at - line 6.
-Character class syntax [= =] is reserved for future extensions at - line 7.
+Character class syntax [. .] is reserved for future extensions at - line 8.
+Character class syntax [= =] is reserved for future extensions at - line 9.
+Character class syntax [: :] belongs inside character classes at - line 10.
+Character class [:zog:] unknown at - line 19.