On Tue, Feb 10, 2004 at 05:02:04PM -0500, George Schlossnagle wrote:
| >
| >What do you think of the new Mail::SRS algorithm?
|
| How does this reconcile the reverse-path size limitations imposed by
| rfc2821?
|
In theory, if we want to be entirely respectful of the 64 char limit, a
client could fall back when it encounters a strict server:
C: MAIL
FROM:<srs0+veryveryveryveryveryveryveryveryveryveryvery-long-localpart(_at_)sender(_dot_)com>
S: 550 sorry bub, localpart too long, try again
C *thinks*
C: MAIL FROM:<SHORT-DATABASE-KEY(_at_)sender(_dot_)com>
S: 250 Ok
Mail::SRS has support for both the super-long version and the
database-key version.
In practice, the srs0/srs1 reduction algorithm allows us to keep the
localpart length reasonably bounded. It may not fall within 64 all the
time but at least it won't grow without bound.
I think the 64 char limit needs to go away.