W3C home > Mailing lists > Public > w3c-wai-ig@w3.org > January to March 2002

RE: Client Side vs. Server Side Image Maps

From: Al Gilman <asgilman@iamdigex.net>
Date: Fri, 01 Feb 2002 14:26:22 -0500
Message-Id: <200202011926.OAA340396@smtp2.mail.iamworld.net>
To: Nick Kew <nick@webthing.com>, Jon Hanna <jon@spinsol.com>
Cc: <w3c-wai-ig@w3.org>
At 02:55 AM 2002-02-01 , Nick Kew wrote:
>On Fri, 1 Feb 2002, Jon Hanna wrote:
>> > Does anyone know why, it is recommended that you use client-side
>> > images maps instead of server-side.
>> Because the client side maps can give the client information about
>> how the purpose of each area (through highlighting the shapes for
>> graphical users, the alt text on <area>s for text users, the title
>> attribute etc.) This cannot be done with serverside maps obviously,
>> as no more information is available than the is given by the image
>> itself.
>That's not strictly true.  You can define a serverside map which
>is equivalent to a clientside one, in the sense that the UA can
>retrieve it and display the options.  Lynx does that.  Of course,
>it still costs an HTTP transaction.
>What the WCAG and other guidelines are rather weaker on is where
>an imagemap doesn't have regions, but is for example a lat/long
>input to a geographic map - which of course can't be done
>clientside without introducing other - more serious - accessibility
>issues.  My own approach and recommendation is to offer alternative
><input>s of type text in this situation.


Yes, it is important to recognize that this case exists, where the function is
a genuine continuous-parameter function.  Map sites are a staple of web usage

There is a collection of techniques that could relate here.

1. The notion that the (0.0) coordinate value in a map request should be
treated as potentially a request for a pointerless access mode.

2. That user agents should produce requests to server side maps with (0,0)
coordinates under suitable circumstances, such as when no pointing device is in

3. When the application, such as a [geographical] map application, is not
simply a short list of discrete hyperlinks, that the server should mount an
alternate dialog usable from the keyboard alone, and provide it or link to it
in the (0,0) response.

This does not mean that the keyboard accessible dialog should not be started
right on the page alongside the map, as Nick suggests.  It just means that
server side maps should always have a (0,0) response and it should always cover
the possibility that the user is keyboard-only.  This response does not have to
assume the (0,0) coordinate value is a conventional signal, but it should cope
with that possibility.

4. That in future the type of map coordinates used in server-side maps be a
nil-able type and that this 'pointerless, please' semantic be attached to a nil
coordinate indication [and free (0,0) for natural use?].

What do people think?

Is this too fussy to ever expect people to use it?

Or is it the kind of "clean practice around the edges" that people would do as
an access accomodation?

Point 4 sounds risky, at least as far as liberating (0,0) from this
connotation.  Rob Lanphier has been eloquent on the QA list recently on the
point that once users of a language have had a feature, it is very hard or ill
advised to actually take it away.

 Re: Conformance and Deprecated Features

>Nick Kew
>Site Valet - the mark of Quality on the Web.
Received on Friday, 1 February 2002 14:27:38 UTC

This archive was generated by hypermail 2.4.0 : Friday, 17 January 2020 20:36:07 UTC