Subject: Question about crypt.c
To: None <email@example.com>
From: John Woods <John.Woods@proteon.com>
Date: 12/22/1994 11:08:39
Having noticed that it changed recently changed, I read it to see what might
have changed (grr, I gotta set up a CVS shadow for sup...), and instead was
struck by the text
* define "MUST_ALIGN" if your compiler cannot load/store
* long integers at arbitrary (e.g. odd) memory locations.
* (Either that or never pass unaligned addresses to des_cipher!)
Surely the vax is not the only permissive architecture (the 386 doesn't
forbid odd accesses, nor does the 680x0 for x > 1). If this is to be a
"better safe than sorry" case, then you might as well remove the test for
"vax" as well (unless someone, somewhere, is trying frantically to crack
passwords on a VAX 11/780 and needs every bit of speed they can wring out
of the blasted thing...).