Re: Working Group Decision on ISSUE-31 / ISSUE-80 validation survey

Aryeh Gregor, Thu, 21 Apr 2011 16:06:53 -0400:
> On Mon, Apr 18, 2011 at 10:10 PM, Leif Halvard Silli wrote:
>> Clearly, MediaWiki could be programmed to fetch @alt text for that
>> image from title of the target page or something?
> 
> My point was that MediaWiki and vBulletin both produce bogus alt text
> for user-uploaded images to ensure they validate.

The "bogus @alt text" resulting from that MediaWiki bug is an empty alt 
inside a link with no other link text. Per HTML4, this is valid. Per 
HTML5 - before the generator exception is applied - it is invalid. 

What needs to be proved is that it would have been better to omit the 
entire @alt. Because such was the claim from the prevailing side of the 
poll.

Another problem for the argument is that, judging form the bug report, 
the example page you pointed to *did* already validate before the bug 
was filed: The MediaWiki bug was filed using arguments about the 
problem of repeated text for *screenreader users*. 
(https://bugzilla.wikimedia.org/show_bug.cgi?id=368#c0) And repeated 
text is valid, even if it problematic.

Too bad then that to screenreader VoiceOver, it would not have mattered 
whether you used an empty @alt or omitted the @alt: both are equally 
wrong. So it is a bad example, if the intention is to prove that the 
desire to validate had negative effect: it was the desire to validate 
that made you repeat the text in the first place.  And, compared with 
the end result, the repetition caused by the desire to validate was 
arguably a lesser evil compared to listening to long URLs being read. 
Thus, the "placebo" alt would, in this case, be the lesser evil.

So what and for whom would be better by omitting the @alt?

Also, those involved in the bug report must be counted as "hand 
authoring" guys. You did not try a quick fix just to validate. It was, 
in effect, some authoring tool vendor representatives who were 
discussing. Whereas the generator exception side of the poll meant that 
while authoring tools developers were excellent, authors who validate 
were sloppy and lazy etc while at the same time also eager to blame the 
tool.

>  I made this point
> only because the chairs said that there was a lack of concrete
> evidence that bogus values had been used in the past or will be used
> in the future, and I knew of concrete evidence that would be useful if
> the issue were reopened. 

Unless I misread, the problem which the chairs discussed was the claim 
that authors took to simple "placebo" @alt text solutions merely to 
validate, and that this had negative effect on users.

> I gave no opinion on the legitimacy of the
> generator exception. 

I'm glad you brought up the examples.

> Whether MediaWiki specifically could do better
> is a question best raised in its bug tracker (feel free to CC me):
> 
> https://bugzilla.wikimedia.org/enter_bug.cgi?product=MediaWiki

Let's see. I did make a few examples of a better solution, so could be.
-- 
Leif Halvard Silli

Received on Thursday, 21 April 2011 21:51:57 UTC