Le mercredi 4 Août 2004 14:15, Marco d'Itri a écrit : > On Aug 02, Rémi Denis-Courmont <remi-usenet1-ms-ipv6@simutrans.fr.st> wrote: > > Eh? I don't think so. gethostname() will often return a result that > > does > > Who cares? Hosts are supposed to know their own hostname. So, you suggest programmers use gethostname() to determine what to send with EHLO? Any SMTP client that try to do so will obviously break on many systems, particularly those with changing IPv4 or IPv6 addresses, where gethostname() will obviously not return the right hostname. What I assume Mozilla's SMTP client does, sounds the most sensible option to me: that is to say, perform a reverse DNS lookup on its own address, use the reverse DNS name if it succeeds, and the IPv6 address if it fails. My point was, not only is it not always possible to know your externally hostname visible hostname, e.g. if reverse DNS does not work as is sometimes unavoidable with IPv6, but you often do not have a hostname at all. So sending your IPv6 address is sometime pretty much your only option. I'm not saying MTAs should stop imposing restrictions on EHLO to decrease spam exposure. It might be OK to require IPv6-enabled MTA to have a working reverse DNS name, but it is not OK -and will probably never be- to require SMTP clients to have one. As such, SMTP clients within an MTA's relaying addresses range should be allowed to specify its IPv6 address with EHLO. -- Rémi Denis-Courmont http://www.simphalempin.com/home/infos/cv.shtml.fr
Attachment:
pgpT2_dAnfg04.pgp
Description: signature