Bug 111975 – Plugins created during user gestures (or soon after) should not be snapshotted

• Component: Plug-ins
• Status: RESOLVED FIXED

[Dean Jackson]

There are many sites which create plugins in response to user actions, such as clicking on a poster frame. We should make sure that we never snapshot in that case. It's not quite so simple though, because sometimes the user action inserts an iframe that loads the plugin, meaning the action and the plugin creation are separate events. The suggested solution is to "bless" all plugins that were created within a few seconds of a user gesture.

[Dean Jackson]

<rdar://problem/13232172>

[Dean Jackson]

Created attachment 192468 [details]
Patch

[WebKit Review Bot]

Comment on attachment 192468 [details]
Patch

Attachment 192468 [details] did not pass chromium-ews (chromium-xvfb):
Output: http://webkit-commit-queue.appspot.com/results/17134207

[WebKit Review Bot]

Comment on attachment 192468 [details]
Patch

Attachment 192468 [details] did not pass cr-linux-debug-ews (chromium-xvfb):
Output: http://webkit-commit-queue.appspot.com/results/17149143

[Peter Beverloo (cr-android ews)]

Comment on attachment 192468 [details]
Patch

Attachment 192468 [details] did not pass cr-android-ews (chromium-android):
Output: http://webkit-commit-queue.appspot.com/results/17169149

[Dean Jackson]

Created attachment 192559 [details]
Patch

[Peter Beverloo (cr-android ews)]

Comment on attachment 192559 [details]
Patch

Attachment 192559 [details] did not pass cr-android-ews (chromium-android):
Output: http://webkit-commit-queue.appspot.com/results/17177331

[Dean Jackson]

Committed r145421: <http://trac.webkit.org/changeset/145421>

[James Robinson]

(In reply to comment #8)
> Committed r145421: <http://trac.webkit.org/changeset/145421>

This doesn't compile on any chromium port (and probably others) since it adds a reference to currentTime() without #include'ing <wtf/CurrentTime.h>.  The EWS bots all pointed out this compile failure - can I ask why you committed anyway without addressing it?

[Tim Horton]

(In reply to comment #9)
> (In reply to comment #8)
> > Committed r145421: <http://trac.webkit.org/changeset/145421>
> 
> This doesn't compile on any chromium port (and probably others) since it adds a reference to currentTime() without #include'ing <wtf/CurrentTime.h>.  The EWS bots all pointed out this compile failure - can I ask why you committed anyway without addressing it?

I'm going to guess he thought that the delta between the two patches would fix the build failures on chromium (it didn't). Even now, the second patch hasn't made it through any of the chromium EWS bots, and is green/yellow across the board.

[Dean Jackson]

(In reply to comment #9)
> (In reply to comment #8)
> > Committed r145421: <http://trac.webkit.org/changeset/145421>
> 
> This doesn't compile on any chromium port (and probably others) since it adds a reference to currentTime() without #include'ing <wtf/CurrentTime.h>.  The EWS bots all pointed out this compile failure - can I ask why you committed anyway without addressing it?

Sorry, I screwed up badly. As Tim suggested, I thought that the issue was it being included from somewhere that hadn't defined currentTime, but I was too stupid to check that it was included when I moved it to the cpp file :(