On Wed, Nov 06, 2024 at 10:25:29AM +0100, Joe Abley wrote:
> On 6 Nov 2024, at 10:17, Otto Moerbeek <[email protected]> wrote:
>
> > I would guess there are many, many cases of applications using glib's
> > getaddrinfo and some other implementations of getaddrinfo sort as
> > well.
>
> If we imagine that the vast majority of cases where people care about any of
> this are, collectively, "the web", then I think there are probably
> vanishingly few cases in which glibc's behaviour is relevant.
>
> If you think "the web" is not the motivation here, it would probably be
> helpful to spell out why you think that.
>
> > The man page of glibc cites RFC 3484 as the reason for the
> > ordering. My expectation would be you have a hard time convincing the
> > glibc people they should ignore that RFC.
>
> I'm not sure why you think that would be sensible. Updating the advice in
> 3484 is a more obvious option.
>
>
> Joe
For the web happy eyeballs is relevant, which alreaty defeats
shuffling of answer records by DNS servers. That leaves the simple,
unsophisticated applications. I don't think all DNS servers should
be burdened to solve the issues of these applications.
Updating 3484 might be possible. Something like: pick a random one if
some of the addreses turn out to be equivalent?
-Otto
_______________________________________________
DNSOP mailing list -- [email protected]
To unsubscribe send an email to [email protected]
