W3C home > Mailing lists > Public > w3c-wai-gl@w3.org > April to June 2001

Re: alt for icons

From: Al Gilman <asgilman@iamdigex.net>
Date: Wed, 02 May 2001 10:22:09 -0400
Message-Id: <200105021416.KAA12734682@smtp2.mail.iamworld.net>
To: A.Flavell@physics.gla.ac.uk, Anne Pemberton <apembert@erols.com>
Cc: WAI Guidelines List <w3c-wai-gl@w3.org>
Let's look a little deeper at what Anne said.  For reference, I use Lynx a
lot.

Text which has an image alternative should not necessarily blend seamlessly
into the context of flowing text.  The image is an encapsulation, something
which is more distinctly set off from its context than are three words
appearing somewhere in the middle of the sentence.

To retain the rhetorical structure in the composite equivalent page with the
ALT text displayed and not the image, should there be something in the text
presentation -- whether style or punctuation -- which retains this
encapsulation, which distinguishes the ALT text as presented from surrounding
text?  This is a property of the pattern of alternative availability and
transformation, and hence is better allocated to be "inserted by the User
Agent" as opposed to "inserted by the Author."

[I am here glossing over the great difficulty of making _small_ changes in
_any_ UI.  I am not sanguine about getting Lynx to adopt this change, even if
we were to consense it was the best plan.  Can always try.]

This does not make the bad examples that Alan cites good.  But good
examples of
alt text would still be good if they were [good].  If the conscientious author
knows that their ALT text will be styled [alt text] in monospaced,
not-otherwise-styled text contexts, they won't be hindered from writing an
effective functional replacement targeted to _that_ form.  

The seeming exception being that you don't wrap a null string in [] just to
show that there was an invisible spacer there.  That truly vanishes.  I guess
the best way to explain this is that the [...] wrapping _is a style_ to set
_something_ off, and if you have _nothing_ you don't need to interrupt the
reader to set _nothing_ off, replacing nothing with something.

Al

Not thatAt 01:44 PM 2001-05-02 +0100, Alan J. Flavell wrote:
>On Wed, 2 May 2001, Anne Pemberton wrote:
>
>>     I haven't looked at something in lynx for a long time, but I thought I
>> remembered that lynx put the alt tag in brackets < > 
>
>Lynx does not put the ALT text in brackets, and as far as I know
>it never has.  And certainly not pointy brackets. 
>
>You might be confusing this with Lynx's habit (variously in different
>versions) of generating a placeholder in square brackets, when the
>author has failed to supply an ALT attribute.  It used to be something
>like [IMAGE] or [LINK] or [USEMAP]; more recent versions insert the
>image filename, producing exciting sites such as this one
>
>   [USEMAP:circle_1.gif] [USEMAP:circle_2.gif] [USEMAP:life_1.gif]
>                                               [USEMAP:circle_3.gif]
>   [USEMAP:life_2.gif]
>   [USEMAP:circle_4.gif]
>   [USEMAP:life_3.gif]
>
>and so on, ad nauseam...
>
>But where an unsympathetic author _has_ provided ALT attributes (and
>even goes so far as to claim compliance with Bobby validation!), we
>get to see this nonsense instead:
>
>      spacer spacer spacer spacer logo
>
>(139 matches for that phrase at Altavista!).
>
>> .... If this is no
>> longer (or never was) the case, then it would be more sensible to change
>> the browsers so that punctuation is added to set off the image text when it
>> is presented visually.
>
>Sorry, I can't agree.  Then the conscientious author would have no way
>to incorporate their desired text string seamlessly into the text-mode
>result, since there would be no way to tell the browser to omit the
>brackets when the author knew they were superfluous.
>
>The solution IMHO is not to take inappropriate ALT text, as was
>exhibited in the samples on which you are commenting, and then to
>forcibly delimit it in the browser with additional punctutation.  The
>solution is to promote the use of appropriate ALT texts, with
>delimiters supplied or not supplied by the author according to the
>desired context.  And put the blame for doing it wrong on those
>authors who are doing it wrong, whether by omitting ALT attributes or
>by using them badly.
>
>> At 05:08 PM 5/1/01 -0700, Adam Victor Reed wrote:
>> >I have an attention deficit, so I often use lynx to avoid distractions.
>> >Alt text appears in place of graphics, so it merges with other text if
>> >not set off with punctuation. It is NOT accessible if it appears as
>> >
>> >    1. Markup tools icon one Markup tools icon two Markup tools icon
>
>The proffered ALT text here is clearly inappropriate, in a way that no
>amount of browser kludging could repair.  Please, let's not make
>things harder for those who are trying to do it right, in an effort to
>mask the faults of those who are doing it wrong.
>
>If you're interested in commenting further on what Lynx does, it
>shouldn't be so hard to get a copy and observe it in action (unless
>your only access is to a Mac, for which there never seems to have been
>a satisfactory port).  Try starting from
<http://lynx.browser.org/>http://lynx.browser.org and
>following the relevant links. Some on-line services which aim to give
>an impression of Lynx's presentation are cited from the Lynx links
>site, but seem to be ancient Lynx versions or no longer accessible.  
>But anyone interested in text-mode browsers should also look at w3m,
>at least.
>
>best regards
>  
Received on Wednesday, 2 May 2001 10:17:00 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Monday, 7 December 2009 10:47:10 GMT