Rev 4718: Switching away from pre-fetching the string_hash function drops hash() time in http://bazaar.launchpad.net/~jameinel/bzr/2.1-memory-consumption

[John Arbash Meinel]

At http://bazaar.launchpad.net/~jameinel/bzr/2.1-memory-consumption

------------------------------------------------------------
revno: 4718
revision-id: john at arbash-meinel.com-20090930173804-3y2bossnkug69k8m
parent: john at arbash-meinel.com-20090930173110-p03w37bugnjkbtao
committer: John Arbash Meinel <john at arbash-meinel.com>
branch nick: 2.1-memory-consumption
timestamp: Wed 2009-09-30 12:38:04 -0500
message:
  Switching away from pre-fetching the string_hash function drops hash() time
  from 13.9ms down to 14.5ms. 4% loss.
  
  Something we may want to revisit, but if this is an opportunity to make a generic
  StaticTuple type, rather than a Keys specific one, it is probably worthwile.
-------------- next part --------------
=== modified file 'bzrlib/_keys_type_c.c'
--- a/bzrlib/_keys_type_c.c	2009-09-30 17:02:00 +0000
+++ b/bzrlib/_keys_type_c.c	2009-09-30 17:38:04 +0000
@@ -248,12 +248,8 @@
 #endif
 	x = 0x345678L;
 	p = self->key_bits;
-    string_hash = PyString_Type.tp_hash;
 	while (--len >= 0) {
-        y = (*p)->ob_shash;
-        if (y == -1) { /* not computed yet */
-            y = string_hash((PyObject *)(*p));
-        }
+        y = Py_TYPE(*p)->tp_hash(*p);
 		if (y == -1) /* failure */
 			return -1;
 		x = (x ^ y) * mult;