Bug 102066 – [Qt] Enable Mutation observer

• Status: RESOLVED
• Resolution: FIXED
• Component: New Bugs

[Csaba Osztrogonác]

It was enabled on all platform - http://trac.webkit.org/changeset/107454, but it didn't 
enable it on Qt's features.pri. It is revealed by http://trac.webkit.org/changeset/134379,
when zillion inspector tests started to fail.

I think we should enable it. Patch is coming soon.

[Csaba Osztrogonác]

Created attachment 173858 [details]
Patch

[Csaba Osztrogonác]

I'm still running all tests to make sure all tests pass after enabling mutation observer.

[Csaba Osztrogonác]

Created attachment 173865 [details]
Patch

[Csaba Osztrogonác]

Comment on attachment 173865 [details]
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=173865&action=review

> LayoutTests/platform/qt-5.0/fast/js/global-constructors-expected.txt:323
> +FAIL WebKitMutationObserver.toString() should be [object WebKitMutationObserverConstructor]. Was [object MutationObserverConstructor].

I have no idea what is it ...

> LayoutTests/platform/qt/fast/js/constructor-length-expected.txt:48
> -FAIL WebKitMutationObserver.length should be 1. Threw exception ReferenceError: Can't find variable: WebKitMutationObserver
> +PASS WebKitMutationObserver.length is 1

It is progression.

[Csaba Osztrogonác]

It is very important, because r134379 made layout testing hyperslow: 24 -> 43 minutes. Should I skip the 48 failing tests until somebody review this one?

[Csaba Osztrogonác]

Created attachment 173868 [details]
Patch

[Csaba Osztrogonác]

(In reply to comment #6)
> Created an attachment (id=173868) [details]
> Patch

I skipped them by http://trac.webkit.org/changeset/134399 to make 
the bots happy and green again. This patch contains the unskip now.

[Csaba Osztrogonác]

Comment on attachment 173868 [details]
Patch

Clearing flags on attachment: 173868

Committed r134401: <http://trac.webkit.org/changeset/134401>

[Csaba Osztrogonác]

All reviewed patches have been landed.  Closing bug.

[Csaba Osztrogonác]

(In reply to comment #4)
> (From update of attachment 173865 [details])
> View in context: https://bugs.webkit.org/attachment.cgi?id=173865&action=review
> 
> > LayoutTests/platform/qt-5.0/fast/js/global-constructors-expected.txt:323
> > +FAIL WebKitMutationObserver.toString() should be [object WebKitMutationObserverConstructor]. Was [object MutationObserverConstructor].
> 
> I have no idea what is it ...

It matches with the platform independent result. The mac,win,gtk result is
"PASS WebKitMutationObserver.toString() is '[object WebKitMutationObserverConstructor]'"

But this test is skipped on Mac,Win and marked as fail on GTK.
I checked the GTK result on the bot and it is same as our.

[Csaba Osztrogonác]

Aha, I got it - http://trac.webkit.org/changeset/122159 -
"Rename WebCore::WebKitMutationObserver to WebCore::MutationObserver"

I'm going to fix it.

[Csaba Osztrogonác]

(In reply to comment #11)
> Aha, I got it - http://trac.webkit.org/changeset/122159 -
> "Rename WebCore::WebKitMutationObserver to WebCore::MutationObserver"

> I'm going to fix it.
Nothing should be fixed now. 
The vendor prefix will be removed once - https://bugs.webkit.org/show_bug.cgi?id=89231

[Csaba Osztrogonác]

It seems it wasn't a good idea to enable mutation observer on Qt,
because it made layout testing so slow:

- 24 mins --> 34-39 mins on our WK1 release tester bots
- 1 hour --> 1.5 hours on our WK2 debug tester bots
- 22 mins --> 28 mins on our WK2 64 bit release tester bot

It seems it is a serious performance regression. :-/
I'll check the WK1/WK2 perf bot results too.