Re: ipv6 broken
On Jan 5th, Robert wrote:
> Funny, I posted the same patch a week ago:
> http://marc.theaimsgroup.com/?l=mutt-dev&m=107278694308624&w=2 ;-)
>
> Unfotunatelly no one cared. Seems most people don't compile with sasl2 stuff.
> :-/
Sorry! I only just read it after seeing your other post..
On Tue, Dec 30, 2003 at 11:18:03AM +0100, Robert Schiele wrote:
>
> What's wrong with size here? ss_len is non-standard and thus not availlable
> on all platforms.
>
> So I propose to go back to size? Do you see a problem with it?
Nope - looks fine to me! I didn't realise even sockaddr_storage would be
non-standard - I just found the low-down on the KAME website:
<quote http://orange.kame.net/dev/cvsweb.cgi/kame/PORTABILITY?rev=1.54>
- RFC2553 defines sockaddr_storage to have __ss_{family,len} member. During
the discusson for the draft on ipngwg, ss_len was proposed and final result
was not very clear. X/Open document mentions ss_len, not __ss_len.
Therefore, you can see both definition on operating system platforms.
The most portable way of dealing with it is to:
(1) have -Dss_len=__ss_len to unify all occurences (including header file)
into __ss_len, or (2) never touch __ss_len. cast to sockaddr * and use
sa_len.
Next version of RFC2553 (RFC2553bis) will define ss_len (not __ss_len),
and be synchronized better with X/Open document.
</quote>
Cheers,
Patrick