Re: Questionable implementation of IMG ALT attribute as tooltips

Jukka Korpela (jkorpela@cc.hut.fi)
Fri, 23 Jan 1998 09:44:50 +0200 (EET)


Date: Fri, 23 Jan 1998 09:44:50 +0200 (EET)
From: Jukka Korpela <jkorpela@cc.hut.fi>
To: www-html@w3.org
In-Reply-To: <SIMEON.9801222127.A@sf106.essex.ac.uk>
Message-ID: <Pine.OSF.3.96.980123085626.13157A-100000@beta.hut.fi>
Subject: Re: Questionable implementation of IMG ALT attribute as tooltips

On Thu, 22 Jan 1998, James Green wrote:

>  *Incorrect usage*:
>    ALT="There would normally be a picture here..."
>  *Correct usage*:
>    ALT="Pic: Abraham Lincoln"
> (IMO)

Since people who _have_ read the specifications (I assume most people
on the list have) have so different interpretations and applications
of the use of ALT, no wonder authors in general are confused.
See <URL:http://d1.ph.gla.ac.uk/%7eflavell/alt/alt-text.html#howlers>.

Is ALT="Pic: Abraham Lincoln" really correct? In which sense does
the text there act as a _replacement_ for the image? Sometimes,
as the last resort, one might write an ALT text which really is
just a description of an image, but I'd recommend using the style
ALT="[Picture of Abraham Lincoln]". But it really says roughly
"sorry I tried to write a replacement text but couldn't".

For instance, consider a biography of Lincoln. The role of a picture
might be just decorative, or orientative; it's sort of normal to
put a picture into a biography. In such a case, using ALT="" is best,
On the other hand, if it's a picture of Lincoln in some particular
situation, carrying a specific _message_, then one should try to
put the message into the ALT text. Perhaps it's a very awkward attempt
to describe a good picture in just a few words, but it might be better
than nothing; after all, it is _intended_ for situations where the
picture is not displayed _at all_. And to take a last example, assume
that one has an index page through which one can access biographies
of famous men, such as <URL:http://www.infidels.org/library/historical/
john_remsburg/six_historic_americans/>. _If_ one decided to use
_pictures_ on such an index page, then if pictures _only_ are used,
ALT="Abraham Lincoln" would be correct. (Naturally, just the name,
with no reference to something being an image.) In the probably
more sensible choice of putting pictures there _alongside with names_,
we are back in the case where ALT="" would be appropriate.

An early (June 1993) draft for HTML, 
<URL:http://www.w3.org/MarkUp/draft-ietf-iiir-html-01.txt>
described ALT as follows:

  ALT                    Optional alternative text as an alternative
                         to the graphics for display in text-only
                         environments.                         

On the other hand, it gave the following example (typos corrected here):

        Warning: <IMG SRC ="triangle.gif" ALT="Warning:"> This must be
        done by a qualified technician.

which suggests to me that they didn't really _mean_ what they wrote.

But the description itself in that early working draft looks much
clearer to me than e.g. HTML 4.0 wording which says "short description"
in one place and "alternate" in another and tells us that the ALT text
is to be used "For user agents that cannot display images, forms,
or applets". (Notice the use of "cannot" instead of "do not".)

The obvious way of solving the identity crisis of ALT is to use
it for replacements _only_ and start using TITLE for a tooltip-like text
when appropriate. (Actually we might also need a way of giving
_technical_ information such as "200K GIF" separately but for the time
being we have to append that to the TITLE information.)

This would require some textual clarifications to HTML specifications...
but am I correct in assuming that there is actually no work in progress
to create a new HTML spec? When HTML 3.2 was approved, there was
information around about Cougar being developed and a couple of working
drafts on various special topics. But now the W3C activity page
<URL:http://www.w3.org/MarkUp/Activity.html> seems to be just saying
how good HTML 4.0 is, with some statements about "what we are doing"
(with no links!), the texts of which presumable predate the approval
of the HTML 4.0 specification.

What I'm asking is whether it makes sense to suggest or discuss
the development of the HTML language, now that almost everyone seems
to believe that XML+CSS is the solution.

Yucca, <URL:http://www.hut.fi/u/jkorpela/>