Re: AuthConfReq: Presentational Markup

Lars Gunther, Sun, 28 Mar 2010 18:37:24 +0200:
> 2010-03-28 14:24, Sam Ruby skrev:
>>> "thou dost not boil a kid in its mother's milk" is expanded
>>> to a rule where no meat may be eaten with any dairy product, or even
>>> from the same plate.
>> 
>> I think that perfectly sums up the situation here.
>> 
>> I have two children.  While I have been tempted on a number of
>> occasions, I can honestly say that I have yet to boil either.  I
>> routinely combine meat and dairy products.  Mmmm, cheeseburger.
  [...]
> The example might be badly chosen, but the principle from
> Maciej still have some validity. 
> 
> Allowing presentational markup is not guaranteed to make sites 
> inaccessible. Heck, most screen readers even have options to ignore 
> tables used for layout. It is however you look at it, a slippery 
> slope.
> 
> If a power user, knowing what he/she is doing, chooses to violate a 
> rule, so be it. But everyone else is best served by simple rules.

Let's compare layout <table>s vs <strike> or <font>:

(1) Layout tables are bad. But we are not living under the threat that 
pages will be stamped as having an error simply because they have a 
table based layout. 

(2) Students *will* sometimes validate a table based layout and ask: 
Why do you say we shouldn't do it when it validates? But on the whole, 
there is no direct correlation between what validators consider errors 
and what teacher teach out as good practise:

	Example 1: Despite the lack of error messages, the message that layout 
tables are a bad thing, is easy to spread and easy to teach. The issue 
is common sense these days.

	Example 2: "Transitional" doctype flavors dominate over "strict" and 
"frameset" more than 10 to 1". [1]  But the use of CSS still outnumbers 
the use of <font>. And the majority use of <font> is for for visual 
fallback, in tandem with CSS. [2] 

I do think that what constitutes and error can shape what authors do, 
though. But I do also think that validation error messages can lead 
authors to do the third best thing instead of doing the second best 
thing (since the second best thing is forbidden and the best thing 
might not be obvious). Example: Using <span 
style="text-decoration:line-through"> can be worse than using <strike>, 
even in those cases when <del> would have been the most correct thing 
to use. Likewise, using <b> seems to me as better than using <span>, 
even when <strong> would have been  best.

[1] http://dev.opera.com/articles/view/mama-key-findings/#structsize
[2] http://dev.opera.com/articles/view/mama-key-findings/#css
-- 
leif halvard silli

Received on Sunday, 28 March 2010 23:55:36 UTC