W3C home > Mailing lists > Public > w3c-wai-au@w3.org > January to March 1999

Re: Microsoft's response on the ALT attribute and how Frontpage deals with it

From: Brian Kelly <lisbk@ukoln.ac.uk>
Date: Tue, 26 Jan 1999 14:15:41 -0000
Message-ID: <000e01be4936$617260a0$3c92268a@bath.ac.uk>
To: dd@w3.org
Cc: w3c-wai-au@w3.org

----- Original Message -----
From: Daniel Dardailler <danield@w3.org>
To: Brian Kelly <lisbk@ukoln.ac.uk>
Cc: <w3c-wai-au@w3.org>; <w3c-wai-ig@w3.org>
Sent: Tuesday, January 26, 1999 1:55 PM
Subject: Re: Microsoft's response on the ALT attribute and how Frontpage
deals with it

>[I suggest this is the last message on this thread on IG, and this
>should continue on WAI AU group list]
>> To be it seems intuitively wrong for an authoring / conversion tool to
>> away resource metadata.
>This is equally wrong to use the ALT attributes to store metadata that
>is not meant to be there.
>>  I would argue it would be better for the
>> information to be provided and then filtered out if necessary.
>But this would need more convention/syntax for the filtering to work
>(how to recognize this is metadata about the image, not the
>This is not impossible (use ALT with some specific sugar syntax
>recognized by all "FileName=foo.gif; FileSize= 100K" just it hasn't
>happen yet and it's also hackery more than good design (For Metadata,
>look at w3.org/Metadata, and RDF)
>> I'd also argue that the guidelines should at least say that authoring
>> tools should provide options for creating null or automatically generated
>> ALT text, so that an organisation has some choice in this matter (e.g.
>> BBC might choice to provide this information as they have their Betsie
>> for stripping out information).
>I agree ALT="" has some utility (for images with no function, like
>decoration) but content that is meant to be served as is, in HTML,
>should be correct HTML, and ALT=filesize is not right. In your
>example, Betsie is just one way of accessing BBC Online, so the HTML
>has to be correct in all cases.

Hi Daniel
    Thanks for the message.
     On one level I agree with everything you say (i.e. there's a need for
an architecturally coherent metadata framework ).  On the other hand we
sometimes need to be pragmatic.
    Let me give you an example.  One could have a personalised web site in
which access by people with disabilities came through a Betsie-type proxy,
which could strip out confusing information such as file size.   On the
other hand file size metadata could be provided to those who need it.  The
process is not easily reversible.  (Systems designers are likely to come up
with more sophisticated "magic.")
   I would argue that the Page Author guidelines should be pragmatic.  They
can be updated to recommend storing such metadata in RDF - when the RDF
schemas are defined and supporting tools available (which could well be
several years away).
     My concrete suggestion is that:

2.5.8: [Priority 1]
The authoring tool must never insert rule-generated description  text into
the document (default "alt"-text) or a properties field  (place-holder

should say:

 "The authoring tool should normally not insert rule-generated description"

and then explain why, but allow software vendors the options to allow
automated ALT generation.

I think this should be an educational issue rather than a WAI rule (I
suspect that customers of authoring tools may pester software vendors for
this facility).

What do the software vendors think?

Brian Kelly, UK Web Focus
UKOLN, University of Bath, BATH, England, BA2 7AY
Email: b.kelly@ukoln.ac.uk URL: http://www.ukoln.ac.uk/
Homepage: http://www.ukoln.ac.uk/ukoln/staff/b.kelly.html
Phone: 01225 323943 FAX: 01225 826838
Received on Tuesday, 26 January 1999 09:17:26 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 6 January 2015 21:39:42 UTC