W3C home > Mailing lists > Public > w3c-wai-er-ig@w3.org > June 1999

Re: Guideline 1 in The evaluation techniques document

From: Al Gilman <asgilman@iamdigex.net>
Date: Thu, 24 Jun 1999 08:46:23 -0400
Message-Id: <199906241240.IAA171863@relay.interim.iamworld.net>
To: "Bruce Bailey" <bbailey@clark.net>, w3c-wai-er-ig@w3.org
At 07:27 AM 6/24/99 -0400, Bruce Bailey wrote:
>Another reason for the distinction is the (perhaps incorrect) behavior of
>the most popular text-only browser.  Lynx treats them differently.  Missing
>ALT Lynx will treat as [LINK] or [IMAGE].  ALT="" will hide images and links
>from Lynx.  ALT=" " will be rendered as Lynx as [ ].

[good summary; may I offer the following tweaks:]

The hiding of items with ALT="" is under user control; it can be switched off.

Screen reader users require that their systems be able to pronounce
punctuation, but a lot of the time they run with punctuation skipped.
Similarly, the ability to suppress all mention of the most insignificant
images is regarded as a benefit by many users.

>>My major reservation with this guideline is that there may be out there 
>>many people who took the effort to write accessible pages with alt=""
>>where appropriate, and now we tell them to revise their pages, without
>>any reason whatsoever.
>ALT="" is really only an issue for links.  There is agreement that ALT="" is
>often appropriate (and even prefered to some other choices).  The current
>debate is the behavior of authoring agents.  The authoring agents should NOT
>start with ALT="" since this would encourge bad content.  The warnings that
>an authoring agent generates could be more strict than a HTML/WCAG

I agree that warnings can and should be issued without a strict violation
of a WCAG checkpoint.

On the terms of reference of "the current debate" I wonder if you have
drawn such a fine line that you actually managed to miss the mark.  I
believe that the document Chris is maintaining is supposed to serve as a
common reference; both for tools which check existing HTML and for checking
which is performed seamlessly as one creates new material in an authoring
tool.  Even that can be either by writing HTML text or by triggering
generation of HTML text.  In that sense, the debate is not about what the
generator "starts with," but rather what the checks accept, IIRC.

Chris, Daniel: can you clarify this?

Received on Thursday, 24 June 1999 08:40:47 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 6 January 2015 21:01:28 UTC