Revert "Change QWindow::visible when surface visibility in ivi shell changes" 47/8047/2 3.0.0 chinook/3.0.0 chinook_3.0.0
authorTadao Tanikawa <tanikawa.tadao@jp.panasonic.com>
Mon, 2 Jan 2017 16:33:11 +0000 (01:33 +0900)
committerJan-Simon Moeller <jsmoeller@linuxfoundation.org>
Wed, 4 Jan 2017 09:39:18 +0000 (09:39 +0000)
commit31926d7657fc1cc1eb01daa48934b759a51f4616
tree290fb5154f3a2db3a53c0b6473a437cd3c2a8b71
parent7e540a7240a786051f50c1b56f61444628cdc82e
Revert "Change QWindow::visible when surface visibility in ivi shell changes"

This reverts commit d0f95885030e5f425c0423a7a29414a204aaa341.

This is root cause of issue that screen doesn't change
from blank screen after booting.

ivi-surface's visibility must not link to QWindow's visibility
because changing it's visibility to 0 means destroying wayland
surface (--> eglsurface --> ivi-surface) immmediatly.

On the other hand, ivi-surface is created with
visibility == 0. Therefore, creating ivi-surface triggers
self destruction of ivi-surface itself immediately.

And to make matters worse, Qt5.7's hiding(visibility == 0)
window method, wl.attach(nil, 0, 0) was wrong, it cause
unpredictable operation of weston (In our case, it drops some
events between WindowManager and Weston).

For linking ivi-surface's visibility and visibility in QML,
we should consider another method. At least, commit d0f958's
approach is incorrect in Qt5.7 so far.

Change-Id: Ibe90f9dae634aa5cdb3c166caf4db02dfd71965b
Signed-off-by: Tadao Tanikawa <tanikawa.tadao@jp.panasonic.com>
recipes-qt/qt5/qtwayland/0002_expand_ivi-shell.patch [deleted file]
recipes-qt/qt5/qtwayland_%.bbappend