Fix for

Subject: [ID 20000903.001] \w in utf8-strings
Message-Id: <E13VUS5-0000cv-00.pgcc-forever-2000-09-03-09-44-29@fuji>

and various related nits.

p4raw-id: //depot/perl@7030

diff --git a/lib/unicode/mktables.PL b/lib/unicode/mktables.PL
index 608a325..37b6e84 100755 (executable)
--- a/lib/unicode/mktables.PL
+++ b/lib/unicode/mktables.PL
@@ -9,9 +9,9 @@ $PropData = "PropList.txt";
 
 # Note: we try to keep filenames unique within first 8 chars.  Using
 # subdirectories for the following helps.
-mkdir "In", 0777;
-mkdir "Is", 0777;
-mkdir "To", 0777;
+mkdir "In", 0755;
+mkdir "Is", 0755;
+mkdir "To", 0755;
 
 @todo = (
 # typical

diff --git a/regcomp.h b/regcomp.h
index e30e67f..225ff74 100644 (file)
--- a/regcomp.h
+++ b/regcomp.h
@@ -192,13 +192,13 @@ struct regnode_charclass_class {
 /* Should be synchronized with a table in regprop() */
 /* 2n should pair with 2n+1 */
 
-#define ANYOF_ALNUM     0      /* \w, utf8::IsWord, isALNUM() */
+#define ANYOF_ALNUM     0      /* \w, PL_utf8_alnum, utf8::IsWord, ALNUM */
 #define ANYOF_NALNUM    1
 #define ANYOF_SPACE     2      /* \s */
 #define ANYOF_NSPACE    3
 #define ANYOF_DIGIT     4
 #define ANYOF_NDIGIT    5
-#define ANYOF_ALNUMC    6      /* isalnum(3), utf8::IsAlnum, isALNUMC() */
+#define ANYOF_ALNUMC    6      /* isalnum(3), utf8::IsAlnum, ALNUMC */
 #define ANYOF_NALNUMC   7
 #define ANYOF_ALPHA     8
 #define ANYOF_NALPHA    9
@@ -220,8 +220,8 @@ struct regnode_charclass_class {
 #define ANYOF_NXDIGIT  25
 #define ANYOF_PSXSPC   26      /* POSIX space: \s plus the vertical tab */
 #define ANYOF_NPSXSPC  27
-#define ANYOF_BLANK    28
-#define ANYOF_NBLANK   29      /* GNU extension: space and tab */
+#define ANYOF_BLANK    28      /* GNU extension: space and tab */
+#define ANYOF_NBLANK   29
 
 #define ANYOF_MAX      32
 

diff --git a/regexec.c b/regexec.c
index 2004cc4..6401710 100644 (file)
--- a/regexec.c
+++ b/regexec.c
@@ -3735,4 +3735,3 @@ restore_pos(pTHXo_ void *arg)
        PL_curpm = PL_reg_oldcurpm;
     }  
 }
-

diff --git a/utf8.c b/utf8.c
index d57b527..a9600e9 100644 (file)
--- a/utf8.c
+++ b/utf8.c
@@ -645,7 +645,10 @@ Perl_is_utf8_alnum(pTHX_ U8 *p)
     if (!is_utf8_char(p))
        return FALSE;
     if (!PL_utf8_alnum)
-       PL_utf8_alnum = swash_init("utf8", "IsAlnum", &PL_sv_undef, 0, 0);
+       /* NOTE: "IsWord", not "IsAlnum", since Alnum is a true
+        * descendant of isalnum(3), in other words, it doesn't
+        * contain the '_'. --jhi */
+       PL_utf8_alnum = swash_init("utf8", "IsWord", &PL_sv_undef, 0, 0);
     return swash_fetch(PL_utf8_alnum, p);
 /*    return *p == '_' || is_utf8_alpha(p) || is_utf8_digit(p); */
 #ifdef SURPRISINGLY_SLOWER  /* probably because alpha is usually true */