W3C home > Mailing lists > Public > public-fx@w3.org > July to September 2012

Re: CSS Masking

From: Brian Birtles <bbirtles@mozilla.com>
Date: Mon, 13 Aug 2012 13:50:27 +0900
Message-ID: <50288793.80009@mozilla.com>
To: Dirk Schulze <dschulze@adobe.com>
CC: "public-fx@w3.org" <public-fx@w3.org>
(2012/08/13 12:26), Dirk Schulze wrote:
> * <FuncIRI> vs <image>
> I understand your concerns but actually I don't see a big problem in that. <image> has support for media fragments and we discussed earlier on CSS3 Image that they don't influence each other

I was under the impression that it's important to be able to distinguish 
between these two uses (i.e. process using the usual CSS image rendering 
logic vs special SVG element handling) at parse time.

We now support both of the following:

mask: url(file.svg#maskElem)
mask: url(file.svg)

And the behaviour is very different. Also,

mask: url(file.svg#rectElem)

Is invalid. If you want to do that, you use element(file.svg#rectElem)

I think that parsing section I wrote an issue about is necessary to make 
sure that url(file.svg#rectElem) becomes invalid and so we don't make 
UAs look up the target element before deciding whether they can process 
as a CSS image or not ?

> * Reference paint servers as mask source
> I would rather avoid special casing mask source and work together with CSS4 Image to support SVG paint servers as input there. CSS Masking is not just for SVG, but for HTML content as well. Therefore I would investigate in a common solution. CSS3 Images started that, hopefully CSS4 Images goes further.

Sounds good. As pointed out to us by Cameron on IRC, CSS4 Images support 
SVG paint servers. However, as we discussed, it seems it does not 
support external ones of the form, url(abc.svg#gradient). So that's 
something we need to look into.

> * <child-selector> = select(<compound-selector>#)
> Why is that a comma separated list of selectors? Can that lead to a selection of multiple mask references?

As discussed on IRC, this let's you do, e.g. 'mask: select(mask, 
linearGradient)'. If we have multiple matches we take the first match in 
tree order (as with document.querySelector).

>> According to that grammar "none alpha" is acceptable. Do we really want
>> to allow that?
> I don't see a problem with it. No mask is applied so.

It was just something Tab pointed out to me in an earlier attempt of 
mine at this grammar. I'm not sure how much it matters?

Thanks again Dirk!

Brian
Received on Monday, 13 August 2012 04:50:58 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Monday, 13 August 2012 04:50:58 GMT