Re: Towards a new CSS Validator?

Here are my few comments :

Needed functions :

1.       Detect bad selectors (a%b)

2.       Detect errors in property values (color: 3px)

3.       Detect unknown properties + differentiate

·         Known but not valid in the current scheme (border-image in CSS 2)

·         Known but proprietary (filter)

·         Unknown but prefixed (-ms-interpolation-mode)

·         Unknown (pading)

4.       Show multiple version of the CSS when a hack is used (which value is used by which browser ?)

5.       Detect when a rule is not possible

·         Due to HTML model (a > a; p > div; .)

·         Due to CSS logic (a:not(a))

6.       Be very fast and customizable

7.       Propose some refactoring

a % b {

    background-image: 3px;

    background-color: #aaFFFFFF;

    background-color: rgba(66%,100%,100%,100%);

    *background-color: transparent;

    *filter: progid:DXImageTransform.Microsoft.Gradient(

        GradientType=0,

        StartColorStr='#aaFFFFFF',

        EndColorStr='#aaFFFFFF'

    );

}

 

img:not(img) {

    -ms-interpolation-mode: bicubic;

    border: none; #

    position: relative;

}

 

ð  Became (in better formatted (colours: red, orange, yellow, green) and with a log of errors, similar as you can see in attachment)

a % b { // bad selector

    background-image: 3px; // 3px is not a valid value for this property

    background-color: 

OPERA 9- : due to non-support of rgba

#aaFFFFFF; 

FIREFOX 1.5- : due to non-support of rgba and #aaRRGGBB

     transparent;


IE 8- : due to non-conformance with * and non-support of rgba and #aaRRGGBB

transparent;

OTHERS : 

rgba(66%,100%,100%,100%); // not supported in CSS 2.1, but only in CSS 3.0, consider revising your stylesheet scheme

    filter: 

IE 8- : 

progid:DXImageTransform.Microsoft.Gradient(

        GradientType=0,

        StartColorStr='#aaFFFFFF',

        EndColorStr='#aaFFFFFF'

); // a value cannot contains any ',', '=', .

}

 

img:not(img) { // impossible selector

    -ms-interpolation-mode: bicubic; // prepositioned unknown property

    border: none; # // # is not allowed here, it's probably a hack for Safari 2-

    position: 

            SAFARI 2- : due to the failing #

                  (undefined)

            OTHERS :

relative;

}



 

Fremy

--------------------------------------------------
From: "Daniel Glazman" <daniel.glazman@disruptive-innovations.com>
Sent: Tuesday, November 11, 2008 5:35 PM
To: <www-style@w3.org>
Subject: Towards a new CSS Validator?

> 
> I urge all readers of this mailing-list to read Olivier Théreaux's
> last blog article about the fate of the CSS Validator. Read, understand,
> comment. Please. Thanks.
> 
> http://www.w3.org/QA/2008/11/dreamimg_a_new_css_validator.html
> 
> </Daniel>
> --
> W3C CSS WG, Co-chair
> 
>