Index: ChangeLog
===================================================================
RCS file: /cvsroot/tcl/tcl/ChangeLog,v
retrieving revision 1.367
diff -u -r1.367 ChangeLog
--- ChangeLog 2001/02/01 00:58:45 1.367
+++ ChangeLog 2001/02/15 17:09:10
@@ -1,3 +1,12 @@
+2001-02-15 Donal K. Fellows <[email protected]>
+
+ * generic/tclCmdMZ.c (Tcl_SplitObjCmd): Improved efficiency of
+ splitting strings into individual characters by adding hash so
+ that only one Tcl_Obj per character is created. Improves
+ performance of splitting of short strings and makes a huge
+ difference to splitting of long strings, such as is done in the
+ mime package in tcllib. [Bug #131523]
+
2001-01-31 Don Porter <[email protected]>
* win/makefile.vc (install-libraries): Corrected misdirected
install directory for the msgcat 1.2 package.
Index: generic/tclCmdMZ.c
===================================================================
RCS file: /cvsroot/tcl/tcl/generic/tclCmdMZ.c,v
retrieving revision 1.30
diff -u -r1.30 tclCmdMZ.c
--- generic/tclCmdMZ.c 2000/09/20 01:50:38 1.30
+++ generic/tclCmdMZ.c 2001/02/15 17:09:11
@@ -939,15 +939,34 @@
* Do nothing.
*/
} else if (splitCharLen == 0) {
+ Tcl_HashTable charReuseTable;
+ Tcl_HashEntry *hPtr;
+ int isNew;
+
/*
* Handle the special case of splitting on every character.
+ *
+ * Uses a hash table to ensure that each kind of character has
+ * only one Tcl_Obj instance (multiply-referenced) in the
+ * final list. This is a *major* win when splitting on a long
+ * string (especially in the megabyte range!) - DKF
*/
+ Tcl_InitHashTable(&charReuseTable, TCL_ONE_WORD_KEYS);
for ( ; string < end; string += len) {
len = Tcl_UtfToUniChar(string, &ch);
- objPtr = Tcl_NewStringObj(string, len);
+ /* Assume Tcl_UniChar is an integral type... */
+ hPtr = Tcl_CreateHashEntry(&charReuseTable, (char*)0 + ch, &isNew);
+ if (isNew) {
+ objPtr = Tcl_NewStringObj(string, len);
+ /* Don't need to fiddle with refcount... */
+ Tcl_SetHashValue(hPtr, (ClientData) objPtr);
+ } else {
+ objPtr = (Tcl_Obj*) Tcl_GetHashValue(hPtr);
+ }
Tcl_ListObjAppendElement(NULL, listPtr, objPtr);
}
+ Tcl_DeleteHashTable(&charReuseTable);
} else {
char *element, *p, *splitEnd;
int splitLen;
