Suggestion: ALTHREF attribute

Jukka Korpela (jkorpela@cc.hut.fi)
Thu, 15 Jan 1998 11:19:58 +0200 (EET)


Date: Thu, 15 Jan 1998 11:19:58 +0200 (EET)
From: Jukka Korpela <jkorpela@cc.hut.fi>
To: www-html@w3.org
Message-ID: <Pine.OSF.3.96.980115102959.16523B-100000@torvi.hut.fi>
Subject: Suggestion: ALTHREF attribute

(I know that this topic is a part of a much wider problem
of addressing on the Web. But I think some interim solution would be
greatly needed.)

Problem description:
There is currently no way to provide an alternative resource
to be used when the resource specified in the HREF attribute
of an A element is inaccessible (temporarily, e.g. due to
server being offline, or permanently, e.g. due to resource
having been deleted).

Notice that in this discussion, "alternative resource" essentially
means 'alternative _copy_ of a document'. Existence of such
alternatives might be based on simple copying "by hand" or on
automatic mirroring.

Importance of the problem:
Great, since resources are so often moved or deleted and since
a connection can be down or much too slow at times. Even if a document
or collection of document is made available on several servers using
various techniques in order to make it efficiently accessible worldwide,
in _HTML_ one currently needs to link to it using _one_ address.
This makes links very vulnerable.

Discussion of solutions:
An obvious solution would be to change the syntax of the HREF attribute
values so that a _list_ of URIs is allowed. The problem with this is
that links using the extended syntax would not work _at all_ on many
currently popular browsers.
Adding a new (naturally optional) attribute, which might be called
ALTHREF for example, would not have such a disastrous effect. Browsers
can be reasonably expected to ignore unknown attributes. The new
attribute value could be a list (in preference order) of URIs.

As regards to the separator used in the URI list, the comma would
be the intuitively most obvious. But since commas may appear
unencoded in a URI, one needs something else. Probably the space
character would be suitable. (It would have the minor drawback that some
nonconforming documents may use unencoded spaces in URIs.)

Minimal support to ALTHREF would mean the following: When a link is
selected, then the URI in the HREF value is tried first. If access
fails (the concept of failure might perhaps be defined exactly in
terms of HTTP here), then the browser should try the URIs in the
ALTHREF value in succession, possibly after asking and getting permission
from the user to do so. Browsers might additionally provide a method
of following a link so that a menu, consisting of the URIs in the
HREF and ALTHREF values, is presented and the user may select one of them.
(This could be useful when the user can decide, on the basis of the URIs
themselves, which of them is probably the most accessible for him.)

A simple example of intended use is in
http://www.hut.fi/u/jkorpela/test/alturl.html

Yucca, http://www.hut.fi/u/jkorpela/