From: Murray S. K. <ms...@se...> - 2007-03-15 17:54:28
|
On Thu, 15 Mar 2007, Mark Martinec wrote: > No, the bug (appears to be of the same sort) is present in both dk and > in dkim milters. I only reported it for dk, as I didn't know until > yesterday (noticed by SM) that dkim-milter can fall on the same test, > but in a slightly different way. Actually, I just figured out that your original report had to do with signing, but I interpreted it as an issue with verifying, so I had been looking at the wrong code paths. When verifying, the comparison is always correct (at least in libdk) for reasons that are too detailed to bother with here. I hadn't looked at signing. I can see what's going on now and will post patches for both shortly. |