Re: [CSS21] Proof of CSS 2.1 / CSS 2.0

On Wed, 24 Aug 2005, Karl Dubost wrote:
> 
> 2. CSS doesn't need versioning because it's a full set of features 
> remaining compatible. (without incompatibilities in time).

That's the intention, yes -- that a real stylesheet (whenever it is 
written) will work with any real UA.

We have changed the spec in some cases to make this possible, due to 
wide-spread (but consistent) bugs, or wide-spread (but consistent) 
authoring mistakes.

However this isn't why we don't need versioning. "We don't need explicit 
in-file versioning" is merely a statement describing the fact that there 
have not been any real needs put forward that would require versioning.


> > CSS doesn't need versioning because there is no real use case for 
> > versioning in CSS. (At least, I haven't yet seen one.) If you know of 
> > a use case, please explain it (and specifically, explain what the 
> > conformance criteria related to that versioning scheme would be).
> 
> I will come with a few cases today. The CSS validator is one of them, 
> but I guess it will be withdrawn by the WG as a real use case.

Validators by definition have to not trust anything in the file. If you 
put versioning information in the file, the validator can't trust it. (For 
instance, if you have a company policy that you must use CSS 2.1, and use 
the validator to check it, it wouldn't catch a file that said it used 
CSS3.)

Also, authors will want to check stylesheets against multiple profiles.


> Calling a term obsolete means that it has not to be supported anymore. 
> It doesn't exist from an implementation point of view.

We can do this without versioning. Indeed, versioning has no relevance 
here that I can see. (HTML is an example of this.)


> > > 3. CSS 2.1 will drop some features of CSS 2 which are not implemented.
> > 
> > Some features, such as display:marker, have been removed, yes. This 
> > has no effect on either implementors, authors, or users, since the 
> > features were never implemented and never used.
> 
> Just note, that it's the "proof" that I am requesting.

I don't see how doing this benefits anyone.


> > There were many fixes to CSS2, yes. CSS 2.1 is a comprehensive 
> > revision that fixes literally hundreds (if not thousands) of issues 
> > that have been raised over the years.
> 
> OK so the CSS WG admits it's a complete different thing. Which was not 
> the initial comment about versioning, yet the group is saying that no 
> versioning has to be put in place.

Indeed. There is no need for versioning when it comes to CSS 2.1, since 
all the changes were made specifically to cater for the real world.


> Yes each time, you drop something from a specification it's good to see 
> why.

It's interesting. But I don't think satisfying interest is worth months of 
work.


> What are the reason behind the choice. Also when something is dropped 
> and reappear in a later version (ooops not it's not version, how do I 
> call that?)

It's a "version" in the English sense.


> > > I'm a bit confused. There's something not logical. I may have missed 
> > > a step.
> > 
> > Hopefully the above clarifies the situation. Please explain what it is 
> > that you do not think follows, if not.
> 
> You are saying in mail that the technology is heavily versioned but that 
> versioning mechanism is useless.

I guess I don't understand what you mean by "versioned" then.


> > > But the thread tends to confirm that the other issue about 
> > > versioning that the CDF WG raised, and that I have raised is 
> > > important.
> > 
> > Did you reply to Bert's answer to your issue? I did not see a 
> > response.
> 
> Not yet :)) on my todo pile.

I would recommend replying to Bert rather than me; he covered pretty much 
every important reason.

-- 
Ian Hickson               U+1047E                )\._.,--....,'``.    fL
http://ln.hixie.ch/       U+263A                /,   _.. \   _\  ;`._ ,.
Things that are impossible just take longer.   `._.-(,_..'--(,_..'`-.;.'

Received on Wednesday, 24 August 2005 15:56:36 UTC