W3C home > Mailing lists > Public > public-html@w3.org > August 2007

Fwd: Baby Steps or Backwards Steps?

From: Jon Barnett <jonbarnett@gmail.com>
Date: Thu, 16 Aug 2007 13:41:51 -0500
Message-ID: <bde87dd20708161141x7fa9d5e5tdf1af1c03159651@mail.gmail.com>
To: public-html <public-html@w3.org>

On 8/15/07, Gregory J. Rosmaita <oedipus@hicom.net> wrote:

> In certain rare cases, the image is simply a critical part of the content,
> and there is no alternative text available. This could be the case, for
> instance, in a photo gallery, where a user has uploaded 3000 photos
> from a vacation trip, without providing any descriptions of the images.
> The images are the whole point of the pages containing them.
> In such cases, the alt attribute must be omitted.
> unquote
> MUST be omitted?  you're telling us that a "critical part of the
> content" MUST NOT have alt text defined for it?

Absolutely.  If an image IS the content - e.g., a photo gallery, a
vista of the grand canyon - That image CAN NOT be placed with a
textual alternative without changing the meaning of the document.
Therefore, a textual alternative MUST NOT be defined - if it were,
user agents could replace the image with the textual alternative
without noting that an image was replaced, the meaning of the document
would be changed, and users would be confused.  If authors want to
provide a textual *description* of the image, they should use the
title attribute instead.   If authors want to provide an alternate
version of the image, they should use OBJECT instead, which has such
fallback capability.

> quote
> A screenshot in a gallery of screenshots for a new OS:
> <figure>
> <img src="KDE%20Light%20desktop.png">
> <legend>Screenshot of a KDE desktop.</legend>
> </figure>
> unquote
> so, the draft is claiming that a "screenshot of a KDE desktop" is
> enough to provide me with an equivalent user experience to that of
> someone who can perceive and visually process the "screenshot of a
> KDE desktop"?

If you can write some text that could serve as a complete textual
alternative (not a textual description) of the KDE desktop, you may do
so.  Otherwise, you MUST omit @alt, or you will confuse users.
<legend> is not an alternative, it's a supplemental description.

> the "legend" is so vague as to be completely
> meaningless -- if i, as a blind user, do not have an understanding
> of a KDE desktop, how am i supposed to communicate problems/issues
> to a sighted colleague?

Since the <legend> itself is vague, then @alt MUST be omitted so the
user agent knows to inform a blind user that there is an image here
they can't see.  If @alt was provided, but left blank, the user agent
would replace the image with blank text, pretending the image isn't
there.  The blind user would be confused.  If @alt was provided, but
only described the image, the blind user would not understand what is
being described, and would not understand that there is an image
missing.  If you want a textual *description* use @title or another
method (such as a very long <legend>, or accompanying text)

>  i need to know not just that there is a
> screenshot, but what that screen shot illustrates, in as much
> detail as possible -- there's no law saying that a user must
> read such a detailed description in full, BUT it must be fully
> described for those who need a detailed description, and that is
> the author's duty, and the markup's function -- give me a terse
> description of the image, as you have done with the "legend" defined
> for the "figure" but that terse description is ABSOLUTELY meaningless
> unless it is described in detail...

And unless you're willing to craftily write text that serves as a
drop-in ALTERNATIVE to the image, then you must omit @alt.

> neither the HTML5 draft -- nor any other associated document -- should
> instruct authors that alt text is unnecessary in ANY circumstance, let
> alone in the straw-man examples cited in the above-referenced section,
> to which your own "rules" don't apply

Those aren't straw men.  The flickr example is a good one.  You can
*describe* images in a photo gallery, but you often can not (or are
not willing to) provide an actual alternative to images in a photo
gallery.  Pictures that are worth a thousand words can't hardly have
textual alternatives.

When prompted, your typical flickr user would probably leave the
alternate text blank.  He would leave alternate text blank, which
leads to <img alt="">, which leads to a UA replacing the image with ""
(an empty string), because the alternative representation of the image
is an empty string, which leads to a blind user who doesn't know that
an image is missing.

> -- show me the machine that can,
> heuristically, dependably, and in a human-comprehensible form, interpret
> a screenshot of a KDE desktop...

Show me a human that can provide text that can serve as an apt
alternative for a vista of the grand canyon.

Jon Barnett
Received on Thursday, 16 August 2007 18:41:56 UTC

This archive was generated by hypermail 2.3.1 : Thursday, 29 October 2015 10:15:25 UTC