Re: Multiple Content-Location headers

At 21.20 +0000 98-01-13, Ben Laurie wrote:
> It seems to me that in at least some cases it is obvious which object is
> primary. For example, I give Apache-SSL to both Oxford and Cambridge
> Universities, who then publish it on their FTP sites. These are then
> mirrored by various other sites. The primary version is the one that I
> generated (which, unfortunately for this discussion, doesn't have a URL
> at all, at least, not one I'm aware of, certainly not one that is
> accessible by the general public), and all the others are secondary,
> right?

MHTML was mainly meant for the case where two different URLs actually
get to the same physical file. But it could, of course, also be used
in the case of mirrors. This might be very useful, to send an object
with a list of mirror URLs. Sometimes, the sender knows which is
the primary, and could then indicate this. But in many cases, the
poor mail client which produces the message will have no idea which
of several URLs referring to the same physical file is the primary
one.

At 14.15 -0800 98-01-13, David W. Morris wrote:
> The point is that a single resource may exist in english and french
> versions (or an image in high, medium, and low resolution, etc.) and
> based on other information, primarily HTTP header content such as
> Accept-language, the server picks one of the versions to return as the
> response. The server notifies the client and any intervening caches
> that the identity of the response is some other URL via the
> Content-location field. In other words, the versions are not the
> exact same object.

In that case, there would be no need for multiple Content-Location headers.
So that would not be any problem.

------------------------------------------------------------------------
Jacob Palme <jpalme@dsv.su.se> (Stockholm University and KTH)
for more info see URL: http://www.dsv.su.se/~jpalme

Received on Wednesday, 14 January 1998 12:42:41 UTC