W3C home > Mailing lists > Public > www-validator-css@w3.org > January 2011

Re: Validator yields error when type="text/css" not specified.

From: Jukka K. Korpela <jkorpela@cs.tut.fi>
Date: Tue, 18 Jan 2011 13:09:21 +0200
Message-ID: <57792E284E514EBEA28A53F816ED5BDF@JukanPC>
To: "Proton Zero" <proton.zero@gmail.com>, <www-validator-css@w3.org>
Proton Zero wrote:

> Whenever the CSS validator uses CSS inside <style> tags (as opposed to
> giving the validator raw CSS data), if I do not specify
> type="text/css" in the <style> tag, I receive an error: "Unrecognized
> link element or xml-stylesheet PI."

I suppose that you mean that if you use a <style> element without the type 
attribute, in an HTML document that you submit to the W3C CSS Validator, 
then that Validator issues the error message instead of processing the 
stylesheet.

The reason is that when processing an HTML document, the W3C CSS Validator 
performs somewhat Draconian HTML parsing. Formally, it is correct, since 
that the type attribute is required in a <style> element, according to 
published HTML specifications, and without it its content is to be assumed 
to be in an unknown language.

On the practical, it would hardly do any harm to imply text/css, as browsers 
do. There is practicelly no other style sheet language used inside <style> 
elements, so the type attribute is pure formality.

> Granted, XHTML5's specification is not yet complete,

There is no XHTML5. You mean the XHTML serialization of HTML5, which is work 
in progress (optimistically speaking), expected to be completed around year 
2020 or so.

> but it does say
> that if a "type" attribute for the <style> tag is not given, then it
> defaults to "text/css", so the two cases should yield an identical
> result.

The current HTML5 drafts say so, and this will hardly change. But the reason 
why <style> without the type attribute should be recognized and processed is 
that it is common to omit that attribute and we can then know with 
approximately 100.00% certainty that the content is meant to be in (some 
flavor of) CSS.

I'd be grateful if people who work with the W3C CSS Validator could comment 
on this. If the Validator can be changed in this respect, there goes 
(probably) the last reason to use the redundant type="text/css" attribute.

As an author, just add the attribute. It does no harm either, if you type it 
correctly. If you mistype it e.g. as type="text/ccs", well, that's an 
entirely different matter...

-- 
Yucca, http://www.cs.tut.fi/~jkorpela/ 
Received on Tuesday, 18 January 2011 11:10:25 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Wednesday, 27 June 2012 00:14:26 GMT