Bug: 6628376
The issue here is that contentX & contentY can both be 0 if we are scrolling
slowly enough as they are adjusted by the page's scale and can thus round down.
However, this would result in us falling out of layer drag mode even though
we haven't tried to scroll past the edge of the layer. Detect this case, and
stay in layer scroll mode.
Change-Id: I3c655d0d03e8f89887abbe718bd24699c133ee1a
Bug: 6569073
Only nativeDestroy and nativeStopGL need to be called on the UI thread,
so split up destroyImpl into 3 functions, and only have the native
destroy be pushed to the UI thread if necessary. Also make the work that
is delayed be static without references to the finalizing WebView to allow
it to be fully deleted immediately after finalization.
Change-Id: I4e424051e69df0bc409af95ca3f3d2b9e58a6b75
On a page load, the (viewport) metadata settings for the main frame are
reset. Reloading a page with viewport meta tag specifying target density
dpi will result in 2 back to back update zoom density messages to be
sent from WebViewCore to UI thread.
In setupViewport(), remove update density messages in flight. If the new
density is indeed different from WebView's default scale, queue another
message.
Bug: 6509512
Change-Id: I195e40277bf1bcccb0e61d84ed84db1e34556243
Rather than caching the WebViewDatabaseClassic instance in a member,
grab the instance when we need it.
In the case of WebSyncManager - neither it or it's subclass uses
the WebViewDatabaseClassic (and it's subclass is final) so remove
it's initialisation. We must leave the member declared however
to satisfy the published API.
Bug: 6482144
Change-Id: Ibf5186a2851d17f5ad62b35ad543aeadff5618d9
Functor attach should always be successful, but adding a fallback just in
case. Also invalidates the WebView on initial content arriving.
bug:6511995
Change-Id: Ibca16505afec9f693ea4a7cc4966cd6d7353725c
Loop is trigger by a seek to 0 when ended on native side but there is no play
call. So on java side, we detect this and call into native side to trigger a
play after completion.
This fixed the UI problem and keep in sync with the native mode.
Beyond that, we don't need to reload for looping and we don't have the seek
to play artifacts.
bug:5461143
webkit change:
https://android-git.corp.google.com/g/#/c/193750/
Change-Id: I779f3e1fbc789832a1a99d1f17823db6b57b35df
Bug 5993716
Use setCustomSelectActionModeCallback rather than
setLongClickListener to disallow custom action block so
that long press can bring up the paste window.
Change-Id: I916e77dcea7914c02191f0ecda37cd126318807d
Add a setting on WebViewCore to control whether we monitor
the responsiveness of the WebCore thread. Default is not
to monitor.
Bug: 6447214
Change-Id: Ia95e5c769d458dcd24ae50660b2f22e93851956f
Bug: 6490959
Fixes the issue where we will show the old tap highlight
if webkit isn't quick enough to respond
Change-Id: Icd9864d276b6ad311e3f3dc4deaa7085e3769006
Bug 6507332
The IME must be informed whenever the selection or composing
region changes so that it can adjust its replacement regions
and suggestions.
Change-Id: I484b112a2fede6528b0bc506711284b59bd886d1
Make the WebCoreWatchdog aware of the WebViews it is monitoring
(rather than the Activity context which may become stale) and
ensure that the code for the prompt dialog is run on the UI
thread.
Bug: 6420310
Change-Id: Ied003938edb04858c85bcc2491c4b2c4c0ede6eb
