On 3 Apr 2006, at 16:17, Michele Bini wrote:
> Ok, thank you; I'll send the new patch as soon as I'm done.

Here you are the new patch; I tested it again.

BTW, I previously did ran the right tests on osx, too. The previous
patch worked by chance as I'm using the flags value only to send it
back in the type3 message and the server is ignoring it, relying on
the fact I'm accepting those it sent in the type2 message. Also, I
think it didn't crash due to the fact that osx has a default
unaligned memory access handler, performing correct (and slow) byte
accesses in these cases.

Anyway, sorry again,
Michele