- From: Dirk Schulze <dschulze@adobe.com>
- Date: Thu, 4 Dec 2014 12:53:00 +0000
- To: Erik Dahlström <ed@opera.com>
- CC: Elliott Sprehn <esprehn@gmail.com>, Rik Cabanier <cabanier@gmail.com>, "public-fx@w3.org" <public-fx@w3.org>, Philip Rogers <pdr@chromium.org>
On Dec 4, 2014, at 12:39 PM, Erik Dahlström <ed@opera.com> wrote: > On Thu, 04 Dec 2014 03:54:19 +0100, Rik Cabanier <cabanier@gmail.com> wrote: > >> On Tue, Dec 2, 2014 at 11:24 AM, Elliott Sprehn <esprehn@gmail.com> wrote: >> >>> If all the isolation property does is create a stacking context [1][2] >>> then it seems like it should be called stacking-context: true to reveal >>> it's purpose, >>> >> >> Its purpose is not to create stacking context. It's designed to limit the >> backdrop for its children with blending. >> The fact that the spec says to do this using a stacking context, is for >> implementors; not authors. >> >> As Erik Dahlström noted, this property also applies to SVG which has no >> stacking contexts. [1] >> >> >>> otherwise we're just going to have blog posts about the "secret css hacks" >>> to create stacking contexts using isolation: isolate as stacking contexts >>> have all kinds of other side effects. >>> >> >> How would this be different from "will-change: transform;"? >> That creates a stacking context with the same side effects. >> >> >>> The property also does not seem to be specific to blending, and the >>> isolation naming is confusing given that there's talk of layout/style >>> isolation, bidi isolation, and now blend isolation. >>> >> >> It's meant to be used with blending and filters but as with many other >> properties, it has side effects. >> >> The next level of the spec will also reintroduce support for non-isolated >> blending. Since this is expensive, authors will be able to opt into this >> with this same property. Non-isolated blending will not introduce a >> stacking context. >> >> I agree that the name is somewhat confusing. We (= mailing list + css >> group) went over different options a couple of years ago and this was the >> one that we eventually settled on. >> >> 1: >> https://groups.google.com/a/chromium.org/d/msg/blink-dev/WoLwgoPB-GE/LITzZ2ifVVsJ > > Was having a 'blend-' prefix ever discussed (as in: blend-isolation)? I couldn't find any mentions of it when searching through the w3 mailinglists. > > Would 'blend-isolation' be an acceptable new name? ‘isolation' will likely isolate for compositing as well. Compositing is not part of level 1 but will be added into future specs eventually. This is one reason why we decided to have the mix- prefix for blending. I agree with Rik here that the purpose of 'isolation' is not related to having a stacking context. It may cause the creation of a stacking context just like filter, transform, opacity and many other properties may do. As a note: We have two more implementations beside Blink which support the ‘isolation’ property. One (Safari/WebKit) is shipping with it in a release version already and another (Firefox) is about to ship in the stable branch soon. IMO this is the worst timing to change names or even functionality. Greetings, Dirk > > > -- > Erik Dahlstrom, Web Technology Developer, Opera Software > Co-Chair, W3C SVG Working Group >
Received on Thursday, 4 December 2014 12:53:31 UTC