Yadis Spec .9 Change Proposal: Sect 6.2.6

Josh Hoyt josh at janrain.com
Tue Feb 7 00:35:00 UTC 2006

On 2/6/06, Joaquin Miller <joaquin at netmesh.us> wrote:
> >We feel strongly that the header if present should always be used,
> >regardless if there is a <meta> and/or the <meta> value differs.
> Thanks, Larry.  I like this argument.  Does anyone have a good
> argument for either of the other two alternatives (priority to the
> meta element or don't care)?

The only "benefit" to "don't care" that I can possibly think of is
that sloppy implementations may be in spec (that is, implementation is
1% easier). It may be a real problem for some users if different
consumers treat that situation differently. So I'd be surprised if
there is anyone who is more than -1 on that.

There is a small positive for the <meta> getting priority, which is
that it makes it possible for a user to override the server settings.
This is probably a pretty rare situation (you get control over the
HTML head, but not over the headers, and the server sets the YADIS
header), and it has the (pretty big) drawback that it makes it
necessary to parse the HTML EVERY TIME that the page is fetched
(effectively making HEAD requests forbidden).

Anyway I just thought I'd flesh out why I *keep saying* that the
X-YADIS-Location header should always win. If there are any arguments
that I missed at all, feel free to add, but I hope that we can finally
put this point to bed.


More information about the yadis mailing list