[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: Length Limit for display-name
> On Sat, Sep 18, 2010 at 8:51 AM, Ned Freed <ned.freed@xxxxxxxxxxx> wrote:
> >> Section 3.4. Address Specification of RFC2822 specifies
> >> name-addr = [display-name] angle-addr
> >> What are the length limits on the display-name field? I take it a long
> >> display-name
> >> value would simply be folded as described under 2.2.3. Long Header Fields
> > AFAICT the relevant specifications impose no limits on the size of the
> > display-name. What implementations actually do is another matter, of course. I
> > believe I've encountered limits as low as 64 octets. I certainly wouldn't
> > recommend putting in stuff longer than 100 or so octets and expect it to pass
> > through everything intact.
> Thanks a lot for your replies, Ned and Hector!
> Indeed, I tracked down the display-name rule in
> display-name = phrase
> phrase = 1*word / obs-phrase
> where 1*word implies no length limit (RFC 2234)
> When I asked about clients, I meant clients that connect to the mail
> server to send an email message, not end-user
> or display clients.
THat excludes the thnings that impose by far the shortest limits on these
> For example, qmail-remote, and gateways when
> passing a message to another server. Here
> I thought the To: header being within the message they would not touch
> it, or parse it and just pass it on, so
> that "trace fields are prepended to the message" (A.4 in the Appendix )
Dream on. That's not how things work in th real world. Lots of intermediaries
perform all sorts of processing of header fields. The legality or illegality of
doing this is mostly irrelevant, since you have to deal with what's out there,
not what the standards-writers hoped would be out there, but even then,
specific header modifications are specifically condoned by vaiorus standards,
e.g., MIME downgrading.
In the specific case of To: fields, an obvious one is for submission servers to
rewrite addresses to eliminae short form names and other local crap.
> Would such clients leave a display-name unchanged?
The answer is no, you cannot count on it.
> I would think only
> the total email length limit would come
> into play, which is usually pretty large (measured in MB)
Again, you're dreaming if you think this is true.