Re: css compliance

[Frank Lin PIAT <fpiat@klabs.be>]

On Tue, 2008-11-25 at 11:58 +0100, Pallinger Péter wrote:
> Frank Lin PIAT írta:
> > On Fri, 2008-11-21 at 12:52 +0100, Pallinger Péter wrote:
> >> The page is not css-compliant, despite the logo. In particular, -moz-border-radius is
> >> not a valid css attribute.
> > 
> > We have this kind of reports from time to time, but that is a false
> > negative.
> > http://lists.debian.org/debian-www/2007/10/msg00283.html
> > 
> > Quoting Richard Atterer:
> > The validator is wrong, the property "-moz-border-radius" is permitted
> > by the spec:
> > <http://www.w3.org/TR/2003/WD-css3-syntax-20030813/#vendor-specific>
> > Maybe the validator classifies it as an error because the use of vendor
> > extensions is discouraged.
> 
> Actually, this is not a false negative.
> 
> According to W3C:
>   "The specification defines a mechanism by which vendor specific
> extensions won't clash with future standard properties, but it doesn't
> define the extensions, and they are not part of CSS 2.1."

You are not quoting W3C, but a post on a mailing list, so I won't
comment it.

> And:
>   "A valid CSS 2.1 style sheet must be written according to the
> grammar of CSS 2.1. Furthermore, it must contain only at-rules,
> property names, and property values defined in this specification"

The above paragraph is really weird, as it conflict with the "-moz-foo"
vendor extension (the CSS 3.0 drafts seems to have dropped that
paragraph actually).

Anyway, the CSS specification explicitly allow "-vendor-foo" for
vendor-specific *extensions*. It does explain how a conforming user
agent should handle it (i.e ignore it by default).

> So, these extensions are not "permitted" by the spec

Wrong, they are permitted. The specs states "Authors should avoid
vendor-specific extensions"

> only recommended to be used by vendors for experimentation (and adhere
> to the grammar rules). And this does not make them valid.

Yep.

BTW, a CSS2 stylesheet isn't valid for a CSS1 conforming user-agent (and
so on).

> It is somewhat confusing for me, too, but this seems to be a consistent view of W3C
> guys, so using '-moz-border-radius' really makes a stylesheet not valid CSS2.1 or CSS3.

Well... so it isn't part of the specs, but it is allowed by the specs.
This is a vendor *extension* that "Authors should avoid". 

I am pretty sure that the author of the stylesheet was aware of it, and
he/she decided to "experiment" that feature (assuming the extension
would be adopted in CSS3??)

You are welcome to provide some patch that implements a similar feature
in a portable and standards conforming way.

Thanks,

Franklin