Thorns pruned - thanks! [long]

Thanks to all for a lot of food for thought!  This is a compound reply to
about 11 of the replies to the original post.

> Steve writes:
> > One of our accessibility-course delegates ... wants to represent greek
letters
> > on their page.

From: <kynn-eda@idyllmtn.com>

> Why? To what end? That is the question you always need to start with.
> What is the purpose of these greek letters?
>
> > They currently use <FONT> to switch to the symbol font
> > to do this, and another delegate argues for Gifs with ALT text.  My
> > views are:
> > 1) Font tags
> > Are wrong.
>
> They may be wrong, but why are they wrong?

Bong!  Once again Steve leaves out the all-important 'context' part of a
communication!  I should have mentioned (a) the delegates actually want to
represent mathematical and chemical formulae, both large, laid out ones and
simpler ones in-line.  They are after sum and product symbols too, as well
as hebrew characters such as Aleph. (b) The university (my client) has a
policy to achieve WAI-AA status on all its pages.  While I believe the
guidelines are set to change any moment, for now I am taking a literal
interpretation of the university's requirements.

> Would it be just as wrong
> to use <span> and CSS to set the font?

Given the above, no, because FONT tags are precluded by WAI guidelines,
whereas SPAN tags are permitted.

> Steve, out of curiousity -- you say <font> tags are wrong but your email
was
> full of <font>. Is there a reason?

Yes, I'm using a recently-reinstalled Outlook Express and was set to HTML
mode.  When I say something is 'wrong' I mean in the context of some goal,
in this instance, the university's achieving WAI-AA status.  I myself am not
an activist: I am a passivist who prefers to let the technology correct my
behaviour.  Of course I am from time to time forced to confront some issues
for pragmatic reasons.  (BTW I'm set to plain text now, hope that helps!)


From: "Lloyd G. Rasmussen" <lras@loc.gov>

> I would suggest using character entities.  Results will vary, but many
> screen reader/browser combinations can pronounce the Greek character
names,
> especially if you move through that portion of the text one character at a
> time, and you have properly declared the charset of the HTML page.
>

You say 'many' but I tried this on the university's systems: IE 5.5 on win98
rendered most characters, ICAB and IE 5 on the Mac did well, but NS, Opera
and Lynx did not.  NS output the entity names which is probably incorrect
behaviour but on a practical note was better because at least it said Theta;
instead of just drawing an empty glyph or no glyph at all.

Just to double check, I declared the charset utf-8, is that right?

Here's the page for those who want to try rendering the entities in their
own browser: http://www-cgi.york.ac.uk/~psyc9/cgi-bin/entities.pl



From: "Jon Hanna" <jon@spinsol.com>

> If you do use entities be aware that some of the mathematical symbols
> are not the same entities as the Greek letters that share the same
> glyph, e.g. &Sigma; is a capital sigma, but &sum; is a sum, even
> though they normally look the same.

Oo thanks, I hadn't noticed that.

> In theory at least (and perhaps someone can fill us in on the
> practice) screen readers could pronounce &Sigma; "sigma" or "capital
> sigma" if standing alone, or as part of a Greek word. &sum; would be
> pronounced "sum of" or "sum".

Yes please, does anyone know what the current state of the art is out there?
I am trying to balance two aspects on this course (a) the university's goal
of WAI-AA status (b) actual accessibility.  If &sum; is the way to go from
the POV pure/correct HTML, then all well and good, but if it crashes screen
readers then maybe a lesser solution like an IMG is better for the meantime.



From: "Al Gilman" <asgilman@iamdigex.net>

> For one thing, character entities are usable in attribute values.  So, if
you are targeting
> level three browsers or some such legacy window, you can do things like
>
> <IMG TITLE="Gamma: CRT response exponent"
> ALT="&#[something]; (Gamma)" SRC="/images/fonts/13-point/gamma.gif"
> CLASS="symbol math colorimetry">

Interesting approach, thanks.

> If the person in your group is working with a class in an organized
institutional setting,
> they should look into getting members of the class to upgrade their
browser to something
> that supports MathML and take the high-tech route.  Because they probably
want to go
> for expressions and equations, too, and not just a few symbols here and
there.

What browsers do support MathML?  This might be a reasonable approach in the
setting we have.  The university also wants to represent math on public
pages, e.g. in the departmental prospectus, but I guess you could argue
that, well yes it's inconvenient, but it's reasonable to expect someone who
is aiming to become a mathematician to download something that will read out
math to them.

> GIFs and ALTs are still a practical approach in some contexts.  Consult
the EASI and
> MAVIS pages for more.

Thanks for these, I will duly peruse.



From: "David Woolley" <david@djwhome.demon.co.uk>

> For heavy maths, I think I would suggest the current best approach
> would be to author in TeX  or eqn/troff formats and provide both
> source and PDF versions.  These are mark up languages that are designed
> for maths, if reasonably laid out, will degrade to plain text in the
> same way that HTML does (HTML seems to have been influenced by them
> in some ways).

Hmm so you are saying ignore HTML, lay it out 'for paper' but with
linearization in mind.  I find this solution initially appealing because it
is close to what they already do.  (On my other wing, I am fighting the
culture the delegates are working in: they do not have a lot of time to
dedicate to accessibility and they do have an awful lot of material up).  My
contact at the university will be upset because it will look like I am
saying 'don't bother with web accessibility'.  But the guidelines say 'use
W3C technology where approriate, and you could argue that it's not
appropriate until MathML is widely supported in user agents.

> If HTML is mandated, for heavy maths, I would suggest capturing an
> image of the relevant part of the PDF page and including the TeX or
> eqn source as alt text, although one might consider using the Greek
> entities, rather than the original names for those characters.

Another interesting one.  I think I will put these to the delegates and see
which they think will be the best way for them to work.

> I suspect that, even if Math-ML gets widespread support in visual
> browsers, it will not be supported in assistive technology except as
> raw XML.  While it may make machine processing and scaleable visual
> presentation easier, it may not have much impact on accessiblity.

Does MathML encapsulate the semantic meaning of an expression?  Or is it a
spatial layout language?  We need the former.


From: "Charles McCathieNevile" <charles@w3.org>

> Actually if you are looking for MathML support it is there for using with
> assistive technologies now. (At the moment via plugins ather than standard
in
> IE, but there are a lot of things IE isn't, and only a few things it is -
> just as you don't use Word for reading the Web, it doesn't mean much that
IE
> only reads HTML.
>
> Have a look at the MathML pages, they have a fairly comprehensive
> implementation report for MathML 2.0.

I will.  Perhaps it will answer a few of the other questions raised by this
discussions.  Thanks everyone.

Steve.

Received on Monday, 11 February 2002 06:17:34 UTC