- From: Sue Sims <sue@css.nu>
- Date: Sun, 09 May 1999 22:16:26 GMT
- To: www-style@w3.org
In discussion on the news:comp.infosystems.www.authoring.stylesheets newsgroup, the W3C CSS2 'Validator', and the WDG csscheck have recently come under scrutiny. A long time user of both these tools, I've recently found myself adding caveats when referring to error messages or warnings issued by each. Since csscheck is a CSS1 lint, it is perfectly reasonable for it to issue an error for inclusion of a CSS2 property. However, that can be confusing for an author not thoroughly conversant with changes/additions from CSS1 to CSS2 (and CSS3, by logical extension). The primary concerns I have with the W3C CSS 'Validator' (apart from the use of the term 'validator'), won't be addressed by this suggestion, but FYI, they are: Warnings about 're-definition' in stylesheets which declare a more specific rule for an identical selector declared previously. The recommendation is specific on the 'cascade'; why would the 'validator' complain about an author's attempt to take advantage of the cascade? Secondly, in the 'validated' css that the tool outputs, any such 're-definitions' are eliminated, *along* with any comments which may have been in the css. If the intention is to 'cut and paste' corrected CSS from the output, removal of the comments lessens the usefulness. The suggestion: Consider allowing specification of the level of CSS *in* the external CSS file, so that CSS lints could operate in much the same manner as HTML validators which use the <!DOCTYPE...> to determine the HTML level against which to validate. That would remove the onus of deciphering error messages from the document author. Something like (as the first entry in an external CSS file): /* url (http://www.w3.org/TR/REC-CSS1) */ Then, it would just be a matter of convincing the lint authors to recognize the reference, and structure their errors/warnings accordingly. There might also be an author who wished to be made aware of *all* potential errors/warnings, so... /* url (http://www.w3.org/TR/REC-CSS1), http://www.w3.org/TR/REC-CSS2/, http://www.w3.org/PR/REC-CSS17/ */ might be useful, as well. Sue Sims
Received on Sunday, 9 May 1999 18:20:10 UTC