W3C home > Mailing lists > Public > www-style@w3.org > August 2005

Re: [CSS21] Lack of version control for content

From: Henri Sivonen <hsivonen@iki.fi>
Date: Tue, 30 Aug 2005 13:01:07 +0300
Message-Id: <bbf94f109e7943401bfd34af90eb2886@iki.fi>
Cc: W3C Style List <www-style@w3.org>
To: Chris Lilley <chris@w3.org>

On Aug 30, 2005, at 04:26, Chris Lilley wrote:

> On Saturday, August 27, 2005, 9:50:48 PM, Henri wrote:
>
> HS> On Aug 25, 2005, at 16:44, Chris Lilley wrote:
>
>>> Experience with the CSS validator shows that the lack of version
>>> identification is a significant problem for managing CSS content; 
>>> there
>>> is no way to indicate to which of the multiple, changing, overlapping
>>> versions of CSS the stye sheet is attempting to comply.
>
> HS> Why is it important for a style sheet to tell what it attempts to
> HS> comply with?
>
> So that it can be validated.

For that it would be sufficient to allow the user of the CSS validator 
to specify the expected profile, which is already allowed. What is 
needed is a CSS 2.1 profile in the menu.

> So that an author can chose which version to output.

The author can choose which properties to use. The choice to produce a 
style sheet according to a profile does not require being able to 
assert the profile in the style sheet itself.

> HS>  Isn't it more interesting to ask "Does this style sheet
> HS> comply with CSS 2.1?"? In that case, it is the user of the 
> validator
> HS> who should give the desired CSS version as additional input.
>
> HS> Compare this with how DTD validation and RELAX NG validation work. 
> I
> HS> think the RELAX NG model with the validation process taking two
> HS> independent inputs (the schema and the document) answers more 
> useful
> HS> question than the DTD model in which the document provides its own
> HS> schema.
>
> An interesting comparison, but flawed. Removing the link to the actual
> schema is a good thing - but CSS has never had such a link. RelaxNG
> schemas can hook into a version attribute, for example - but CSS 
> doesn't
> have one of those either.

The crucial question is why one would want to associate a schema like 
that. It makes sense for autocompletion as an editing aid. It makes no 
sense for applications that want to verify the input conforms to their 
expectations. It makes limited sense for validation.
See http://groups.yahoo.com/group/rng-users/message/52

In my opinion, the question "Is this valid Foo?" has merit. Also, the 
question "What is this?" can be interesting. I think "Is this valid 
whatever it is?" makes the least sense of these three questions as a 
quality assurance question.

In the case "Is this valid CSS 2.1?", the user of the CSS validator 
gives the validator two inputs: the style sheet and the desired CSS 
profile. No need for internal version assertion.

In the case "What profile(s) of CSS does this style sheet conform to?", 
the best answer is given by testing the style sheet against all known 
profiles. (No need for internal version assertion.) A style sheet can 
syntactically conform to multiple profiles of CSS at the same time. The 
answer that a style sheet conforms to CSS1, CSS 2.1, CSS3 and CSS MP at 
the same time is a better answer than that the style sheet itself 
wanted to claim compliance to CSS 2.1.

> If the CSS WG does not consider that validation of CSS stylesheets is
> appropriate, please explain that to the CSS validator folks.

That is different from not wanting to put a version label in the files 
themselves.

> Currently,
> the W3C CSS validator seems to assume CSS 2.0, so it complains about
> orange and fails to complain about properties in CSS2 and not in CSS
> 2.1, for example.

So the validator needs to be updated with a CSS 2.1 profile.

> Similarly, if someone is using CSS MP, the validator woudl fail to warn
> about using properties outside that profile.

Then that person would choose "CSS MP" from the profile popup.

-- 
Henri Sivonen
hsivonen@iki.fi
http://hsivonen.iki.fi/
Received on Tuesday, 30 August 2005 10:01:38 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Monday, 27 April 2009 13:54:40 GMT