Re: Need some help

> Essentially, the proposal is that HTTP should be changed to allow the
> title of an HTML document to be requested and retrieved independently of
> the document itself.


I wonder how this differs from the HEAD request, which returns the
<TITLE> as part of <HEAD>.

> image is made.. I have sometimes wondered whether, by making the kind of
> modification to HTTP that Al has suggested, one would be inadvertently
> sending a signal to content providers that they do not need to be diligent
> in ensuring that the alt attribute is used, since the HTTP change will
> serve as a fall-back mechanism that text-based clients can rely upon in
> any case.

I'm with you, but on a purely OO basis, it's good to have the
description of the data (here an image) stored with the image, and
retrieved from the same source. Maybe this is all too theorical and
ALT and ALTSRC is all we need.
 
> Another issue which is also relevant here, is the practice of writing text
> only pages, in circumstances in which the default (graphical) presentation
> can not be readily modified so as to make it accessible. Firstly, I would
> question whether, given the HTML 4.0 proposals relating to tables and
> forms, there will be much need for "text only" alternative pages in the
> future, particularly once the "screen reader" access paradigm has been
> supplanted by the widespread availability of braille and audio clients
> that have direct access to HTML and CSS content. However, if it is
> considered that there will continue to be a substantial need for "text
> only" pages, then it might be reasonable for a client to be able to
> indicate to the server via HTTP that text only pages are to be provided,
> if available.

My personal opinion, and I'm sure you understand it very well: I think
this is even more dangerous than the other proposal: the whole idea of
SGML is to have one version of the markup and different presentation. 

If we start "advocating" different version of the sources, we're in
for trouble, but I got your argument about what is needed today to
facilitate browsing.

Some kind of new text/html-text-only mime-type is one way to approach
the problem.

Another way to implement text/only I've heard before is the use of
LINK with some REL=text-only keyword, so that once the regular HEAD is
fetched, the user-agent can automatically finds the text-only version.

Received on Wednesday, 9 July 1997 11:19:00 UTC