> I'd argue that any machine parsable format can't be ambiguous by virtue
> of the fact machines parse it. However in any case AtkText /
> IAccessibleText / the mac accessible protocol thing all expect the text
> for an object to be a string so whatever format the web uses screen
> readers will be handling a serialized format.
>
Sorry I was not clear here. Concretely, if "{x+1}/2" is read "x plus one over
2", it's ambiguous since that could also mean "x + 1/2". Blind people have
standard rules to clarify reading of mathematical expressions and screen
readers must follow these rules. That's not necessarily the order people
usually read a formula. And reading "opening brace x plus closing brace slash
two" is not what they want either. So at some point, someone will have to parse
and understand the mathematical structure. That's essentially what MathPlayer
and a MathJax extension prototype do from the MathML tree. I guess at the end
they will send text strings to the reader anyway.
> > - For people with reading disabilities (dyslexia etc), you need to
> > synchronize highlight of equation parts / reading of equation parts.
> this should actually work quiet naturally in my proposal since we can
> tell API consumers that the bounds of characters in the serialized text
> are those for the formatted text shown on screen. That doesn't quiet
> work for { / } / ^ etc, but we can just give them a size of zero or
> something, and that should probably be fine.
It seems that you are oversimplifying the issue here. First people proposing an
alternative syntax will still need to define a mapping from the text source to
the visual 2D representation if you want to know which part to highlight. For
MathML it's the normal mapping from DOM to the rendering tree in Gecko but it's
not clear what Benoit wants to use instead since without further info it seems
that the DOM will just have a text node. Next, what you suggest seem to only
work for variable names and it's not clear how you'll read e.g. operators or
square roots. IIRC, to read a fraction the tools I mention highlight the
numerator when they read it, then the whole fraction's bounding box when they
read the "over", then the denominator when they read it. Or perhaps they
highlight the whole fraction before, I don't remember exactly but the point is
that it's not limited to text. I admit that I don't know the details but there
are other similar sync rules between highlight and reading
. Also, I hope you realize that mathematical layout is much more complex than
just fractions and scripts so I don't think one can just say "I guess that will
work" for such a nontrivial problem.
_______________________________________________
dev-platform mailing list
dev-platform@lists.mozilla.org
https://lists.mozilla.org/listinfo/dev-platform
