Remove the SVG instance tree

• Component: SVG
• Status: RESOLVED FIXED

[Philip Rogers]

Supporting the instance tree requires a lot of complexity and it can probably be removed without affecting any users.

The instance tree was ahead of it's time (many of the same issues are on public-webapps right now!) but it never caught on with users. Chromium's usecounters found no usage of instanceRoot and it was removed entirely in mid-2014[1]. I'm not aware of any web compatibility fallout from removing it (not a single bug, AFAIK). There is support in the SVGWG for removing the instance tree in SVG2[2]. The instance tree was never implemented in Gecko.

Removing the instance tree in blink took a few patches but most were mechanical. All the patches are listed on crbug.com/313438 and would be pretty easy to port to WebKit.

[1] https://groups.google.com/a/chromium.org/d/msg/blink-dev/mgYCJuzfU0Q/8jdhjY0pNgoJ
[2] https://lists.w3.org/Archives/Public/www-svg/2014Jan/0014.html

[Darin Adler]

Sounds good to me.

[Sam Weinig]

I'm fine with this as well.

[Darin Adler]

I’m going to take a crack at this.

[Darin Adler]

Investigating crbug.com/313438 further, it’s clear this is a pretty big project that Rob Buis did! But we can indeed take advantage of those patches and redo this for WebKit; it will be 10 or so patches to do it all. I am about to land a first patch.

[Darin Adler]

Created attachment 246251 [details]
Patch

[Darin Adler]

Committed r179810: <http://trac.webkit.org/changeset/179810>

[Chris Dumez]

This may have caused a perf regression on the HTML5 full-render test:
- https://perf.webkit.org/#mode=charts&chartList=%5B%5B%22mac-mavericks%22%2C%22Parser%2Fhtml5-full-render%3ATime%22%5D%5D (Mavericks / 3.6% regression)
- https://perf.webkit.org/#mode=charts&chartList=%5B%5B%22efl%22%2C%22Parser%2Fhtml5-full-render%3ATime%22%5D%5D (EFL / 2.3% regression)

I am not 100% sure it is caused by this change but the range on the EFL bot is small enough that I'd say it is very likely.

[Chris Dumez]

(In reply to comment #7)
> This may have caused a perf regression on the HTML5 full-render test:
> -
> https://perf.webkit.org/#mode=charts&chartList=%5B%5B%22mac-
> mavericks%22%2C%22Parser%2Fhtml5-full-render%3ATime%22%5D%5D (Mavericks /
> 3.6% regression)
> -
> https://perf.webkit.org/
> #mode=charts&chartList=%5B%5B%22efl%22%2C%22Parser%2Fhtml5-full-
> render%3ATime%22%5D%5D (EFL / 2.3% regression)
> 
> I am not 100% sure it is caused by this change but the range on the EFL bot
> is small enough that I'd say it is very likely.

Looking at the change though, I really don't see what could have caused a regression. I think Ryosuke is planning to confirm the regression locally (may be a flake on the bot or another commit).

[Darin Adler]

Are there any SVG <use> elements in the full-render test?

[Said Abou-Hallawa]

*** Bug 121050 has been marked as a duplicate of this bug. ***