- From: Yves Lafon <ylafon@w3.org>
- Date: Fri, 14 Mar 2008 06:30:28 -0400 (EDT)
- To: Benjamin Hawkes-Lewis <bhawkeslewis@googlemail.com>
- Cc: www-validator-css@w3.org
On Fri, 14 Mar 2008, Benjamin Hawkes-Lewis wrote: Benjamin, responses inline: > The CSS 2.1 spec defines some rules for parsing malformed declarations: > > http://www.w3.org/TR/CSS21/syndata.html#parsing-errors > > Based on the two CSS grammars and these rules, I would expect that given the > input: > > h1 {color: red;} > p {font-weight: bold; width: 100em !ie; color: green;} > em {color: yellow;} > > the validator would extract the following CSS: > > h1 { > color: red; > } > p { > font-weight: bold; > color : green; > } > em { > color: yellow; > } > > since while width: 100em !ie cannot be tokenized as a CSS 2.1 declaration, I > think it can be tokenized with the core CSS grammar like this: > > (ruleset: (selector: (ident: p) (whitespace) ) ('{') (declaration: > (property: width) (whitespace) (':') (value: (dimension: 100em) (whitespace) > (delim: !) (ident: ie) ) ) ('}') ) > > However, the validator drops the entire ruleset, extracting only: Yes, in case of error, the current behaviour is to skip until a matching brace. This might be changed in the future, but is part of a more general discussion about: "Should a _validator_ do error recovery like other UA like browsers, or not? The same apply for the examples below. In a way yes, if you want to "correct" the CSS and display what a "conformant UA" should read from the malformed CSS, however it's not necessary if the goal is mainly to display errors. > h1 { > color: red; > } > em { > color: yellow; > } > > I noticed that the validator doesn't handle some of the examples of malformed > declarations given in the spec correctly. > > (a) p { color:red; color; color:green } > > > (a) p { color:green; color{;color:maroon} } > > (c) p { color:red; color{;color:maroon}; color:green } > > The validator should have extracted color: red; in (a) and (c) and color: > green; in (a), (b), and (c), but in fact it extracts no valid CSS > information. > > Can anyone confirm if this reflects a bug in the validator? > > -- > Benjamin Hawkes-Lewis > > -- Baroula que barouleras, au tiéu toujou t'entourneras. ~~Yves
Received on Friday, 14 March 2008 10:31:01 UTC