W3C home > Mailing lists > Public > public-html@w3.org > August 2008

Re: Images and alternative text

From: Leif Halvard Silli <lhs@malform.no>
Date: Sat, 09 Aug 2008 01:33:13 +0200
Message-ID: <489CD7B9.9000100@malform.no>
To: Philip Taylor <pjt47@cam.ac.uk>
CC: Dave Singer <singer@apple.com>, "public-html@w3.org" <public-html@w3.org>

Philip Taylor 2008-08-08 20.43:

> Dave Singer wrote:


> It's not fundamentally 
> different to assuming users will recognise and understand what "Σax²" 
> means, except that it's a different choice of syntax.


TeX/LaTeX is more than a different syntax. It also uses other 
symbols than the normal math symbols.

 
> (Some users won't understand one or the other or both of those syntaxes, 
> just like some won't understand the surrounding English prose or the 
> concepts being explained, and that's okay since I would only be writing 
> equations for a specialised audience.)

But if you are user friendly, then you allow me to know that you 
wrote the equation as TeX code. Then I have the chance to look up 
TeX to understand what you meant.

>> Yes, if you want to put Latex in there (or any other text), you'll 
>> need to put something before an initial {;  perhaps something that 
>> helps you diagnose that it is, in fact, latex source, and not really 
>> alt text at all. 


hear hear.

>>  Perhaps, even
>> alt="{latex} {x \over y} = {1 \over {y \over x}}"
>> ?
> 
> According to the current HTML 5 spec (as I understand it), that means 
> the image is a key part of the content, and there is no textual 
> equivalent of the image available, and the kind of image is "latex} {x 
> \over y} = {1 \over {y \over x}".

I agree with Phillip's reading of the current spec.

> I believe the LaTeX code *is* a textual equivalent (because it's text 
> (being a human-readable string of mostly-ASCII characters) and it has 
> equivalent meaning to the rendered image without any complex encoding), 
> so that meaning is inappropriate; and the extracted string between the 
> outermost "{" and "}" is clearly not what was intended, and it's not the 
> kind of image anyway. (The kind would be something like "equation", but 
> that's thoroughly unhelpful to users who can't see the image.)

It is *not* a textual *equivalent*.  It is a textual *fallback*. 
It is a "if you can't understand the image, perhaps you would like 
to read the source code I used to generate it from?"

Of course source code can sometimes be used as fallback.

> Something like alt="LaTeX: {x \over ... }" would avoid those problems, 
> but still that's ugly (particularly if you have many equations in a 
> single sentence) and redundant (since it's pretty clear when text is 
> LaTeX, to readers who understand LaTeX) and it's not entirely obvious 
> how to implement this correctly (as you demonstrated by using "{latex} 
> {...}" which'll get badly misinterpreted :-) ).

One simple rule could be to say that if the <img> has lang="zxx", 
then the @alt="{keyword}" rule won't apply.
-- 
leif halvard silli
Received on Friday, 8 August 2008 23:34:01 UTC

This archive was generated by hypermail 2.3.1 : Monday, 29 September 2014 09:38:57 UTC