Re: Collapsing 0 width margin; CSS version system

> > That is only the case because CSS lacks a standard method of version
> > control (for lack of the proper term) -- the CSS equivalent of a doctype
> > declaration. Isn't that the case?
> 
> Not really. Changing behaviour based on the version is very confusing to
> authors -- just look at the confusion people have over why their pages
> change rendering when viewed in quirks mode vs standards mode, or for
> pages sent as text/html vs pages sent as applicaiton/xhtml+xml.
> 

I agree completely. The versioning model used for HTML works for it
fairly well (and any other XML-based, or XML-like language that might
get grouped together with it) and at least allows for transition of
those specifications without breaking older specs. Once web standards
are the norm and specs (such as XHTML 2) arrive it will be important
to differentiate between XHTML 1 and 2 since both will be useful.

With CSS standing away from any particular language and where the spec
strives to 'improve on' itself rather than replace itself the 'levels'
system is far more suitable than versions in my opinion as well.
 
> > Until such versioning becomes part of the standard, knowledgable web
> > authors will continue to devise and use a wide variety of CSS hacks
> > [1][2][3][4]. Of course these hacks are a major pain in the butt, but
> > they are considered a necessary evil, much like table layout, spacer
> > gifs, and many other workarounds used to be.
> 
> These hacks work around bugs in browsers, not lack of versioning in the
> spec.
> 

Having said all the above... I think partial implementations are an
advantage with CSS, we don't have to wait a long time for more
urgently demanded spec improvements (or at least, we don't now that
there is some browser competition). While advantageous in terms of
availability, partial implementations do cause problems that have led
to the need for browser bug hacks. I'm not going to repeat myself
(I'll link back to my original post on the matter), but I really do
think that the "!required" syntax that was suggested last month would
resolve the above issue and remove the need for hacks over unsupported
properties (albeit not faulty implementations). That message is
archived at: http://lists.w3.org/Archives/Public/www-style/2004Dec/0084.html

I think that while versioning is the wrong way to go, we should make
some effort in removing the dependence on technical browser hacks in
*design* degradation decisions. What if Internet Explorer 7 has no
'easy' unique css hack ("* html", I'm looking at you)? If we have a
!required-like syntax prepared in time for IE7 (an idea which I think
the IE team would entertain, certainly the likes of Dave Massey have a
genuine desire to fix IE, even if not right this second) then we've
got a chance of having CSS3 (and 4 and 5...) implemented with a syntax
that will allow people to move on with new properties at a rate that
suits their project, not browser support. And that's without having to
settle for 'all or nothing' degradation that comes with browser
sniffing and/or the time constraints to produce multiple style sheets
for older browsers - that kind of duplication is surely what we want
to get away from?

Ben

-- 
http://www.ben-ward.co.uk

Received on Tuesday, 11 January 2005 18:40:41 UTC