Don't send any preferences that have a string value that is longer
than MAX_ADVISABLE_PREF_LENGTH. This is intended to mitigate OOM
issues, as I've seen a parent process crash trying to create a 100mb
message to send to the child. Such users likely cannot use e10s at
all.
This has a test for all combinations of setting the default and user
values of a preference to large or small string values, or not setting
them at all.
I manually verified that filtering out preferences reduces the size of
the IPC::Message that is sent to the child by printing out the size of
the reply message in PContentParent::OnMessageReceived().
The list-traversal code would not account for the space after the comma, and
therefore would now disable the MediaWidevineNoWMFNoSilverlight case.
Fixing the list for now, I will fix the code in a later patch.
MozReview-Commit-ID: 9ETL56A9wMu
In bug 1264497 we discovered that Netflix was broken because we'd made a change
to not send the Adobe GMP the device bound nodeId, which it stored in GMP
storage. When the GMP was comparing the nodeId it had stored against what it was
passed (nothing) the comparison was failing. Users could have worked around this
problem by clearing their GMP storage, whereupon the Adobe GMP will have stored
"nothing" as its nodeId.
When bug 1264497 was fixed, users who'd cleared their storage will see Netflix
fail again, as their stored nodeId of "nothing" will again not match what we
pass in. So to fix Netflix for these users, we need to clear GMP storage.
This is another instance of a more general problem that we have occasionally
encountered, namely that sometimes GMP storage becomes incompatible, and we
need to clear it. Having a general mechanism that we can use to clear storage
remotely will be helpful, so this patch adds one, and triggers it to fire.
This mechanism is pref controlled, so that we can issue a hotfix if necessary
to clear GMP storage.
MozReview-Commit-ID: GzSyBj0P2JG
--HG--
extra : rebase_source : b854860ee533b0742a664c862278ce54c9052596
That is, the single caret in cursor mode will always persist on all
platforms as on Firefox Android.
MozReview-Commit-ID: 5MTCf1n2dF3
--HG--
extra : rebase_source : 4062752d7c781acc19088106028e848d1192f880
Add a bool flag in nsJAR to track whether the zip handle came from
omnijar. The old code compared pointers instead, but other patches in
this bug made it no longer possible to do that.
Add a pref "layout.accessiblecaret.always_show_when_scrolling" defaults
to true on all platforms except b2g. When it is set to false, the carets
will be hidden during scrolling, which is the current behavior before
applying this change.
The pref "layout.accessiblecaret.extendedvisibility" was added for
Fennec to keep ActionBar open when carets temporarily hiding during
panning or zooming. Now we make carets always show by default, so the
pref can be removed. However, the floating toolbar still need to be
notified when the scrolling begins, so we dispatch "scroll" instead.
In gtest, the preference changes were in the middle of the test
function. To make the preference change clearer, I add new pref changes
or move the existing ones to the beginning of the test functions.
The 250ms transition effect added in ua.css is per request of UX
designer in bug 1249201 comment 12.
MozReview-Commit-ID: 8NGvDLPbtNY
--HG--
extra : rebase_source : 3f7a9ebdf4c70b0282dbf9e8f18cbe5cca656dbe
Patrick McManus
Andrew McCreight
Gerald Squelart
Chris Peterson
Daniel Holbert
Dan Glastonbury
Dragana Damjanovic
Jonathan Kew
Wes Kocher
Mathieu Leplatre
danhuang1202
Carsten "Tomcat" Book
Chris Pearce
Jocelyn Liu
Kirk Steuber
Ehsan Akhgari
Kyle Huey
Henry Chang
Gabriele Svelto
Xidorn Quan
Matt Woodrow
Yoshi Huang
Andrea Marchesini
Tobias Schneider
Ting-Yu Lin
vinayak
Jim Chen
dimi
Milan Sreckovic
Paul Adenot
Kit Cambridge
Bas Schouten