From: Archie Cobbs (archie@whistle.com)
Date: Tue Dec 22 1998 - 14:14:25 EST
Parmelan, Edouard writes:
> I have some comments/questions about your hashtable implementation:
>
> - If sizeof(void *) > sizeof (long) on a 64 bit hardware, the definition
> of DELETE ((const void *)~0L) could be a legal address (with out
> alignment). If so, you should use a definition like this:
> static const void * DELETED = (const void *)&DELETED;
OK, sounds good.
> - In hashFindSlot(), the assert(0) should be explain. May be with:
> The hashtable is never full because the rehash limit is alway
> less than his size (and never equal).
Yes.
> - In hashFindSlot(), the following patch reduce the collision path
> by saving only first DELETED index.
Yes again :-)
> - I also add utf8const$(OBJEXT) dependencies in kaffeh/Makefile.in
> as I compile kaffe in an other directories tree (better for cvs).
That's already fixed now.
Thanks!
-Archie
___________________________________________________________________________
Archie Cobbs * Whistle Communications, Inc. * http://www.whistle.com
This archive was generated by hypermail 2b29 : Sat Sep 23 2000 - 19:57:27 EDT