Date: Tue, 17 Jul 2001 00:20:31 +0900 From: itojun@iijlab.net To: Ian Dowse <iedowse@maths.tcd.ie> Cc: freebsd-net@FreeBSD.ORG, Bernd Walter <ticso@mail.cicely.de>, Hajimu UMEMOTO <ume@mahoroba.org> Subject: Re: how to get AF_LOCAL from getaddrinfo() Message-ID: <14350.995296831@itojun.org> In-Reply-To: iedowse's message of Mon, 16 Jul 2001 15:25:25 %2B0100. <200107161525.aa15954@salmon.maths.tcd.ie>
next in thread | previous in thread | raw e-mail | index | archive | help
>>I'd very much like to see PF_LOCAL support added to our getaddrinfo()
>>and getnameinfo(). I know that PF_LOCAL sockets have semantics that
>Here is quick and simple implementation - any comments welcome. It
>probably needs a few changes to match the conventions of other
>implementations, and a mention of these conventions on the manpage.
what is the main motive of your change?
your change is correct in terms of address family independence,
however, due to the impact of the change, i would really want you to
be careful as hell. if i were you i won't go there.
(i don't think i will bring the change back to kame tree, i don't want
to affect other platforms)
the change will break a lot of applications. specifically:
- TCP-only clients, including ftp. (*)
- daemons which usees AI_PASSIVE to open listeing sockets. they won't
get broken, but they will have a AF_UNIX socket unwillingly opened
for listening.
(*) you can blame us (or whoever else) for not checking AF_INET/
AF_INET6 on getaddrinfo loop, but anyway, there are a lot of
applications that does not.
also, with AF_UNIX, {NI,AI}_NUMERIC{HOST,SERV} does not have proper
interpretation. you will need to pick some interpretation (maybe
you may want to follow what NRL did for linux glibc/NRL IPv6 stack)
again, if i were you i won't go there.
itojun
To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe freebsd-net" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?14350.995296831>