W3C home > Mailing lists > Public > www-svg@w3.org > July 2003

Re: Help requested on accessibility of SVG

From: Jim Ley <jim@jibbering.com>
Date: Tue, 29 Jul 2003 23:19:27 -0000
To: www-svg@w3.org
Message-ID: <bg79op$mqm$2@main.gmane.org>

"Chris Lilley" <chris@w3.org> wrote in message
> I think the root of your problem is conflating the alternative text
> (Valid W3C HTML 4.01) the displayed text (W3C  HTML 4.01)

I disagree, the image has one meaning "Valid W3C HTML 4.01", and if we're
going to have text in the document, then it should where possible add to the
meaning, consider a stronger case where we put a cross in the document to
denote an invalid document, not having invalid there would mean users can
get very different ideas ("invalid HTML 4.01" and "HTML 4.01" are almost

I agree it would be fine to not have the text as text, but to just use desc
and title to provide the text content (you can still have user selectable
text and cut+paste doing this.) which would allow us to fully specify
sensible phrases as the equivalent for groups.  However that is not the
approach SVG has taken.  Because of the fact that text is text, we have to
make that text meaningful - using symbology for banned/invalid etc. is a
problem since someone looking at only the text part of the document will
miss an important part - consider road-signs such as "Parking" with a line
through it, the TITLE of the image would be "no parking", but the text in
the document would be parking, quite the opposite of the intended meaning. A
user interogating part of the document for the meaning could get easily get
misled, however if the only text was "no parking", and the text in the image
wasn't really text, there's no opportunity for confusion.

The badge example isn't a particulalry good example of the problem, the text
doesn't need to say "valid W3C html  4.01", I still believe it would be
desirable if it could.  Repeating content is not desirable when you could
get it in one.

> It could also be argued that making text display as a symbol (or as
> nothing, which is also possible) is in itself inaccessible.

You could certainly argue that, but that's a big problem for all image
formats, the way SVG has gone is a peculiar one, in that it places text in a
special place in that you can select and copy the meaning of the glyphs out
of the document, however you cannot select and copy the meaning of any other
element out of the document, even if the user specifies it *.  This is
inconsistent and makes it difficult to mix text and symbols within the same
document, where the symbols carry meaning.  selecting everything and copying
out of 2 SVG documents denoting "Parking" and "No Parking" signs would give
you the same content "Parking" yet the documents mean quite the opposite.

Because of that problem, I want to use symbols that can be selected and
copied out as a whole, which carry the same meaning as the document, or at
the very least can not mean the opposite. The current approach of the
altGlyph is a hack, but the alternative (a mechanism in UA's to select and
copy a text equivalent of non-text content) is not available.

> Read the license for Bitstream Vera. You can use the glyphs, but you
> can't call them Bitstream Vera anymore so please edit the font-family
> attribute.

Oops, yep, thanks for pointing it out.

> You do not need to use altglyph to get a W3 ligature.

I just nicked it from Dean.


* A UA could of course choose to do this, but it has to provide the text
Received on Tuesday, 29 July 2003 22:21:06 UTC

This archive was generated by hypermail 2.3.1 : Wednesday, 8 March 2017 09:46:56 UTC