Re: Alt="" or " "?

On Thu, 28 Sep 2000, William Loughborough wrote:

> In summary: Authors must decide: 1) if a particular image is purely 
> decorational; 2) it's an image not worth writing any descriptive or 
> replacement text for;

I've maintained the view that ALT text is meant to be a functional
alternative for the image, not "descriptive".  If something
descriptive is required, you have the TITLE and LONGDESC attributes
available.

The unfortunate spot in the HTML4 specification where ALT is described
as "short description" in my opinion is an editorial blunder, and is
incompatible with the better-informed description at
http://www.w3.org/TR/REC-html40/struct/objects.html#h-13.8 where it
says: "alternate text to serve as content when the element cannot be
rendered normally".

"To serve as content" and "to describe the content" are two quite
different concepts, which are in general incompatible with each other,
which fits very well with the fact that HTML has different attributes
for these purposes.  The TITLE attribute is available for a short
description of the content; the ALT attribute is meant to "serve as
content" i.e to be a functional replacement for the image, not to be a
description of it.

Sometimes the correct functional replacement for an image seems indeed
to be a null string, or some kind of separator.  There have been
detailed arguments (some at the level of HTML technicalities, some
based on the actual behaviour of current browsers) as to whether the
use of nothing more than "white space" in the ALT attribute is
acceptable.  Maybe non-white space should be used, but I don't want to
stall the discussion on that kind of detail.

I have to express my dislike of your phrase "not worth writing".  
There are times, in my view, when the correct functional replacement
is indeed null, and writing it is the correct thing to do, rather than
mere laziness as you seem to imply.

I don't see any objection to having null or only-space ALT attribute
with a meaningful descriptive TITLE, in appropriate situations.  It
closely fits the principles of the (better-informed) parts of the
HTML4 specification, for example. The only practical objection for
that is the lack of support for it in certain popular browser
versions.  But it would be a pity to ruin the documented solution
based only on the inadequacy of some currently-available client
software.

best regards

Received on Friday, 29 September 2000 06:58:39 UTC