Allow for changing size of bracketed regex char class

This commit allows Perl to be compiled with a bitmap size that is larger
than 256.  This bitmap is used to directly look up whether a character
matches or not, without having to do a binary search or hash lookup.  It
might improve the performance for some installations that have a lot of
use of scripts that are above the Latin1 range.

1 files changed, 14 insertions, 1 deletions

diff --git a/regcomp.h b/regcomp.h
index f1f55eb6f1..9b11ce5ee1 100644
--- a/regcomp.h
+++ b/regcomp.h
@@ -184,7 +184,20 @@ struct regnode_2 {
     U16 arg2;
 };
 
-#define NUM_ANYOF_CODE_POINTS   256
+/* This give the number of code points that can be in the bitmap of an ANYOF
+ * node.  The shift number must currently be one of: 8..12.  It can't be less
+ * than 8 (256) because some code relies on it being at least that.  Above 12
+ * (4096), and you start running into warnings that some data structure widths
+ * have been exceeded, though the test suite as of this writing still passes
+ * for up through 16, which is as high as anyone would ever want to go,
+ * encompassing all of the Unicode BMP, and thus including all the economically
+ * important world scripts.  At 12 most of them are: including Arabic,
+ * Cyrillic, Greek, Hebrew, Indian subcontinent, Latin, and Thai; but not Han,
+ * Japanese, nor Korean.  (The regarglen structure in regnodes.h is a U8, and
+ * the trie types TRIEC and AHOCORASICKC are larger than U8 for shift values
+ * below above 12.)  Be sure to benchmark before changing, as larger sizes do
+ * significantly slow down the test suite */
+#define NUM_ANYOF_CODE_POINTS   (1 << 8)
 
 #define ANYOF_BITMAP_SIZE	(NUM_ANYOF_CODE_POINTS / 8)   /* 8 bits/Byte */