|
From: Jonathan D. <im...@ya...> - 2006-06-25 21:27:13
|
Hi Ok, tracked down the problem with Snefru on big-endian machines. It re-ordered the bytes regardless of the native ordering, so on big-endian machines it would totally mangle things. This should now be fixed. I've applied the various submitted patches, cleaned up a whole bunch of redundant code in alder32 (and elsewhere) and tidied up the configure stuff. Byte ordering macros have been provided in mutils, to and from native, for 16-bit and 32-bit words. I've also added memory locking into mutils, so that when libmcrypt is ported over, the memory functions can all be handled there. Still looking at treetiger, as it may be possible to extend the concept to ANY hash algorithm - in which case, this will be an option that is handled by a flag and not as a distinct hash. Patches are very welcome. Patches that add new hashes are VERY welcome. Patches that will help with adding aforementioned tree would also be VERY welcome. Jonathan __________________________________________________ Do You Yahoo!? Tired of spam? Yahoo! Mail has the best spam protection around http://mail.yahoo.com |