Case-insensitive parsing of <id-right> in message-IDs

[Russ Allbery]

Julien ÉLIE <julien at trigofacile.com> writes:

> In lib/hash.c, we have the cipoint() function only used in
> HashMessageID().  The message-ID is then hashed after the
> "case-insensitive point" with lower-case chars.

Yeah, this is silly; the postmaster bit in particular is based on a
strange misreading of the mail standards.  I think there may be something
in TODO or in the bug tracker about this.

> Should we still go on using that "case-insensitive point" or could we
> just drop the use of that function in INN 2.6?

The problem is that all existing INN history databases have the
case-insensitive version of the message-ID hashed, which means that if we
just change this, the history database will no longer match articles with
uppercase characters in the message ID that have already been received and
we'll accept duplicate articles.

I've wanted to remove this for a while, but I could never figure out a
transition plan.

-- 
Russ Allbery (rra at stanford.edu)             <http://www.eyrie.org/~eagle/>

    Please send questions to the list rather than mailing me directly.
     <http://www.eyrie.org/~eagle/faqs/questions.html> explains why.