Re: Proposed changes in warning handling in the CSS validator

On Wed, Dec 13, 2006 at 06:59:20PM +0900, olivier Thereaux wrote:

> * by default, show only "high priority" warnings. Those who care can
> ask for more. (the validator should give a hint that there were low
> priority warnings if they are turned off)

I rather suspect that "off by default" would result in the people who
need it most never turning it on.

> One of the consequences of that is that the (rather unpopular)  
> accessibility-related warnings on color and background color will not  
> be present when running with the "default" validation parameters. I  
> gave it a long thought, and although the "no-color" warnings are well- 
> intentioned, they are imperfect, almost out of scope,

Aren't all the warnings out of scope for a validator? Perhaps renaming
the CSS Validator as the CSS Lint would be a good idea?

> and alienating more people than they are helping.

My impression is that there are three groups of people who have issues
with that warning.

1) People who think that "You have no color" means "You have no value
for the background-color property" instead of "You have not specified
the color property as well as the background-color property".

I think this group can be helped by rephrasing the warnings.

2) People who do not understand the reasons for the warning.

Perhaps this is cause for a QA document that the warning can provide a
link to? The subject has been discussed enough times on the mailing
list that writing such a page should not be too difficult. One of
Jukka's[1] emails (copyright issues aside) could probably be used almost
verbatim.

3) People who have code that throws warnings but don't want to fix
them (usually due to falling into group 2 above, or because they
depend on transparency to show the background image on an element
being the element to which the color property is applied.)

These people don't usually have a problem with seeing the warnings
themselves (although possibly making the switch to turn warnings off
more prominent would be useful), but with their clients (and potential
clients) seeing the warnings and forming a bad impression of the author.

Perhaps this could be addressed by making the "This document validates
as CSS!" message more prominent, and including an introductory
paragraph explaining the difference between warnings and errors to the
warnings section.

[1] I *think* he is responsible for some of the best explanations I've
seen on the subject, if I'm misattributing them, then apologies to the
true author.

-- 
David Dorward                                      http://dorward.me.uk

Received on Wednesday, 13 December 2006 10:30:08 UTC