RE: Warnings with valid background-color

On Fri, 14 Oct 2005, Matt LaPlante wrote:

> I respect your opinion, but I think it reads too much into the purpose of
> the tool.  A validator should be just that: a validator.

The word "validator" is a misnomer anyway, and it is very common to 
confuse markup validation (against a DTD by SGML or XML rules - a limited, 
objectively definable task) with "CSS validation", which is informally 
defined and a moving target (though practically often more useful than 
markup validation, since it is so much easier to get CSS code wrong).

> It's not a "better idea recommender".

If you want it to check just against the specifications (so that
font-size: 1px; color: black; background: black; is reported as 
"correct"), then we need to know which specification should be applied and 
which parts thereof are treated as normative. This is a nontrivial task 
for CSS, to put it mildly.

Moreover, most people _want_ "better ideas" at least if the ideas point 
out design or typing or thought mistakes that they can understand.
(The background-color issue doesn't quite naturally fall into this 
category, because you need to understand CSS well to see what the warning 
is talking about.)

Besides, at http://jigsaw.w3.org/css-validator/validator-uri
you can select Warnings: no warnings, if that's what you want.

> The code as presented is 100% undeniably valid, so I
> don't see on what grounds a validator should have the job of questioning
> someone's technique.

"100% undeniably valid" is a strong statement, especially when the 
specifications have quite a lot of room for interpretation.

"Questioning someone's technique" is an odd name for useful warnings, 
which point out very common mistakes (and some other). Anyone who knows 
the CSS techniques so well that she knows when the warnings can safely be 
ignored can simply ignore them, or even switch them off.

> Just because some may think writing a sentence in all
> caps is bad form, doesn't mean it gives the HTML validator cause to throw
> warnings when someone does this.

We were discussing "CSS validator", not markup validators. Quite apart 
from this, the world would really need a decent HTML checker that not only 
performs markup validation but also runs different pragmatic tests.

> A validator should be concerned strictly with how code follows spec, not 
> how a designer follows "good" or "bad" technique outside of the spec.

If that's your idea, you can use the "CSS Validator" with warnings 
switched off. It will be less useful that way, but it's your choice.

> And if something is fully valid in the specification, even if the
> specification might be considered lacking by some or many, it remains an
> opinion at that point.

Surely. And most people need expert opinions.

-- 
Jukka "Yucca" Korpela, http://www.cs.tut.fi/~jkorpela/

Received on Monday, 17 October 2005 13:47:27 UTC