On Tue, Sep 29, 2015 at 08:32:04AM -0700, SM wrote: > Hi Bruno, > At 03:14 29-09-2015, Bruno Tournay wrote: > >I guess you meant section 2.2.3. :-) > > Yes. :-) > >What I understand: it is recommended to use folding between items, even > >when it is allowed to fold in the middle of an item. > >But from section 3.6.4 it seems that it is not allowed to fold in the > >middle of References item (defined as "msg-id" in ABNF, by reading the > >penultimate paragraph of section 3.6.4). > > My reading of the "msg-id" (Section 3.6.4) is not to fold. > > >So it seems that References header is badly formatted for two reasons: too > >long (998 limit is a MUST, I didn't know that), and folding does not > >happen in an allowed place. I may be wrong (after a first very quick look > >at the RFC). > > Yes. > > >By the way, I've noticed that the 998 limit is respected... if you ignore > >the header name. :-) So it looks like more a simple bug than poor > >consideration for the standard. > > I would list it as a bug. :-) >
That being said, Eric and I think that we should try to remove limit. The reason why it exists is that the RFC clearly stated it and we were a bit concerned about accepting mail that might not pass through relays. In practice, we have bumped it to 4x times the RFC length, we've never had a mail blocked by a hop and people keep complaining that we block. If we remove the limit, the worst that is going to happen is that smtpd will accept a mail that will be block by another hop, resulting in a bounce being generated. I guess we can live with that. It will require a bit of work to get rid of the limit though, it might take some time before we can effectively remove it. There are also some consequences in the message parser and filter path that need to be assessed first. This is where we're going anyways -- Gilles Chehade https://www.poolp.org @poolpOrg -- You received this mail because you are subscribed to [email protected] To unsubscribe, send a mail to: [email protected]
