W3C home > Mailing lists > Public > w3c-wai-gl@w3.org > April to June 1999

Re: techniques doc: character entity references for math symbols

From: Alan J. Flavell <flavell@a5.ph.gla.ac.uk>
Date: Mon, 3 May 1999 21:43:45 +0100 (BST)
To: Nir Dagan <nir@nirdagan.com>
cc: Wendy A Chisholm <chisholm@trace.wisc.edu>, WAI Guidelines List <w3c-wai-gl@w3.org>
Message-ID: <Pine.OSF.4.10.9905032122080.16497-100000@a5.ph.gla.ac.uk>
On Mon, 3 May 1999, Nir Dagan wrote:

> My experience with Netscape4.51 is that is displays correctly a
> great deal of math symbols if 
>  1. they are marked with decimal references
>  2. in addition the page is served with charset=utf-8

This is also my conclusion.  All versions of Netscape are severely
broken in this area, as they seem to behave as if the character
repertoire is pretty much limited to the repertoire defined by the
value of the "charset", and furthermore they don't seem to have
implemented the character entity names.

On the other hand, displaying something like &gamma; unparsed, even if
it isn't very attractive, is more functionally useful than displaying
the wrong character (which happens on some HTML3.2-conforming browser
versions, because they take the numerical value modulo 256), or an
empty box.

> I have heard rumors to this effect for other 4.x versions.

Indeed.  I recommend (for these kind of situations*) to compose the
HTML using only US-ASCII characters, and to represent other characters
by either their entity names (for Latin-1 only) or their numerical
character references.  If the result is advertised as UTF-8 then it is
entirely valid (albeit it provokes one minor bug in MSIE4, see
http://www.hut.fi/u/jkorpela/HTML/chars.html#theway ), and then even
Netscape manages to display it.

*)I mean, obviously this wouldn't be an attractive way to compose
a page that was primarily in Cyrillic, Japanese or whatever.  But if
the page is primarily in a Latin alphabet, it's OK.

> Lynx 2.8.1 supports *all* math symbols in *all* the different
> ways you can think of. 
..
> A possible modification of the guidelines is to use images with
> alt text according to HTML4.0, this way both Lynx and old, buggy,
> and non-standard graphical browsers will display correctly.

That's kind of poetic.

One could also supply some heavyweight javascript to tell the
graphical browsers how to correct their bugs, while Lynx calmly
displayed the standard HTML correctly.

I'm not sure what conclusions to draw from this about accessibility
guidelines; I'm just trying to contribute my observations from trying
to make this stuff work at all, in the available browsers.

Jukka Korpela has some interesting pages on the topic (start at the
URL that I cited), and is kind enough to link also to my own at
http://ppewww.ph.gla.ac.uk/~flavell/charset/

Best regards
Received on Monday, 3 May 1999 16:43:51 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Monday, 7 December 2009 10:46:59 GMT