Re: lynx and image maps

"Daniel W. Connolly" <connolly@beach.w3.org> wrote:
>In message <29462.9602062006@afs.mcc.ac.uk>, lilley writes:
>>
>>Yes, this is a simple addition to make. We have been running such a 
>>modified cgi program since August 95. I have put together a page that 
>>demonstrates this [1] but those of you who are unable to display pages 
>>without images ;-) may prefer to go direct to the explanation [2] which
>>also has links to download the source.
>>
>>I tested this with emacs-w3 (an old version, 2.1.90), lynx (also 
>>old, v2.3), and Netscape 2.0N for X (with inline image loading disabled).
>>
>>
>>[1] http://www.man.ac.uk/CGU/staff/lilley/test/maps.html
>>[2] http://www.man.ac.uk/CGU/staff/lilley/test/textonly.html
>
>
>I think I'd like to see this sort of thing written up as a W3C tech
>report. Title might be "HTML Imagemap Recommendations for Information
>Providers".
>
>	(It would eventually be released as an informaional RFC,
>	but I don't necessarily want to go through the whole IETF
>	review process for 'recommendations' documents. Of course,
>	at the discrtion of whoever writes the draft, it could be
>	an html-wg work product.)
>
>The HTML 2.0 spec specifies a certain syntax, and gives a little bit
>of semantics, but there's a lot missing in the way of recommended
>practices. TimBL used to maintain a style guide, but he hasn't got
>time to keep it up to date.
>
>Right now, I think the world gets referred to the NCSA documentation,
>which hasn't undergone a peer review process (I think?) and hence
>doesn't always address the minority concerns.
>
>W3C's conformance testing efforts[1] are about to heat up, and concise
>specs on such issues just might begin to have a "force of law" behind
>them before you know it.
>
>[1] http://www.w3.org/pub/WWW/Test/

	This is not a "recommendation", but a review of the
situation, from which a recommentation might be developed.

	Server-side imagemaps began as an NCSA innovation in
conjuction with Mosaic, handled via a server script which expects
to receive an ?x,y coordinate pair as the "QUERY_STRING",
corresponding to the screen coordinates of a mouse click, and
mapped to documents which the script returns, in effect as
redirection.  The markup associated with ISMAP lends itself to a
text client, or GUI client with image handling turned off, sending
a request without an appended "QUERY_STRING", causing the script
to return an error message.

	It's a "simple" hack, from the perspective of someone
with programming expertise, to modify such scripts so that they
treat a NULL QUERY_STRING as a request for a document which is
suitable for a text client, or for a GUI client with image
handling turned off.  One can't predict how difficult, or
well-advised, an equivalent hack would be for contemporary
servers which handle imagemap requests via internal code, and it
may not be possible to modify commercial servers distributed as
binaries.  It would be a GOOD IDEA to change the meaning and the
server and/or script response for an imagemap request which lacks
a QUERY_STRING, but this change could take quite some time to
become implemented universally.

	In the meantime, Lynx (and as far as I know, only Lynx)
has adopted the strategy (as of its v2.4) of sending ?0,0 instead
of NULL as the QUERY_STRING for imagemap requests.  At sites
concerned about servicing Lynx (used for a variety of reasons,
e.g., because it is interfaced to a braille transducer), it then
becomes possible to make 0,0 a default region of images, or
actively map it to a default-like file, which is suitable for
text clients.  This approach now requires that GUI clients be
hacked to send ?0,0 when image handling is turned off, or that
the server's script or code be hacked as above, so that either NULL
or ?0,0 are treated as a request for a suitable document.  The
Lynxism worked reasonably well when we implemented it last year,
but as more and more GUI client users habituate to the glitz for
the sighted, and tour the Web with image handling turned off for
greater speed and efficiency, we're basically back to square
zero.

	Any consistent strategy likely to be implemented widely
would be fine.  It's primarily a matter of making developers and
WebMaster responsive to the need.

				Fote

=========================================================================
 Foteos Macrides            Worcester Foundation for Biomedical Research
 MACRIDES@SCI.WFBR.EDU         222 Maple Avenue, Shrewsbury, MA 01545
=========================================================================

Received on Thursday, 8 February 1996 10:58:44 UTC