Re: [Doh] gethostbyname?

[神明達哉 <[e-mail]>]

At Tue, 26 Mar 2019 07:26:49 -0400,
Ben Schwartz <bemasc=[e-mail]> wrote:

> Please avoid casting aspersions on other working group participants, even
> in jest, as it might be misconstrued and cause offense.

Perhaps the wording wasn't very appropriate, but I think it's
generally a good practice to have any IETF work assume the use of
IPv6.  It's also consistent with the overall trend to IPv6-awareness
in the IETF, e.g., as described in
https://www.iab.org/2016/11/07/iab-statement-on-ipv6/

The actual draft text doesn't seem to be so problematic in that sense:

   Browsers which cannot get the IP address(es) of the resolver
   configured by the operating system using APIs are still able to use
   an operating system function such as gethostbyname() or its
   equivalents to convert host names into IP addresses through the stub
   resolver in the operating system on which they are running.

since it only refers to gethostbyname() as an example and also says
"or its equivalents", and I also don't think the author lives in the
previous century given that he was involved in the development of a
modern DNS library; but I don't think changing gethostbyname() to
getaddrinfo() (or at least gethostbyname2()) doesn't do any harm
either, if it's not an improvement.  I've just sent a pull request for
this change on github:
https://github.com/dohwg/draft-ietf-doh-resolver-associated-doh/pull/6

> > I have a silly question. In section 4, the draft goes to the API level
to
> > specify how to resolve a domain name. gethostbyname was deprecated in
favor
> > of getaddrinfo (and therefore IPv6 capability) in 1999 (RFC 2553), and
is
> > now found in all common OS's including Linux, FreeBSD, Windows, MacOSX,
and
> > so on. A reference to gethostbyname in this draft comes across as
ignorant
> > at best, perhaps deliberately so.
> >
> > What do we need to do to drag the authors into the current century?

--
JINMEI, Tatuya