-Fix miscellaneous compile errors
-Move manual tests to manual folder and enable export of autotests
-Added widgets requirement
-Fixed autotests
-Fixed renderer and controller synchronization in QML case
-Treat fallback Mesa as ES2 similar to setting AA_UseSoftwareOpenGL
Change-Id: If6619733725d079e339bef16262e5ea1450ab20f
Reviewed-by: Tomi Korpipää <tomi.korpipaa@theqtcompany.com>
Apparently Qt modules are now required to use Q_SIGNALS and Q_SLOTS
macros instead of 'signals' and 'slots' in headers.
Change-Id: I4140b1fff4386d74bb371176919234366965e887
Reviewed-by: Mika Salmela <mika.salmela@theqtcompany.com>
Primary fix is to not use shared context in qml with WinRT,
as that causes black screen.
Also removed widget apps from build in WinRT.
Change-Id: Id65ce90dbeb80442194ab838952e8cadc91135a6
Reviewed-by: Miikka Heikkinen <miikka.heikkinen@theqtcompany.com>
Change-Id: I453438fee92f18c983c1bef04c68b24a7ff72cf9
Reviewed-by: Tomi Korpipää <tomi.korpipaa@digia.com>
Reviewed-by: Mika Salmela <mika.salmela@theqtcompany.com>
On linux, there seems to be some asynchronous activity
related to window resizing, since calling processEvents() is not
enough to ensure viewport size gets updated in subViews test.
Use QTRY_COMPARE to allow for small delay.
Change-Id: Idae04ff9436c819ef150caa57f6fa988e4011746
Reviewed-by: Tomi Korpipää <tomi.korpipaa@digia.com>
Some properties have different default values, or cannot be
changed at all on ES2.
Change-Id: Ieb3951f5692a645eafab492819ce18de19134335
Change-Id: Ieb3951f5692a645eafab492819ce18de19134335
Reviewed-by: Miikka Heikkinen <miikka.heikkinen@theqtcompany.com>
Also fixed a bug found in testing, and added tests for
optional constructors for already tested classes.
Task-number: QTRD-3368
Change-Id: I2214f28e2c5069ecab422fc6817acb2f0c0b192b
Reviewed-by: Miikka Heikkinen <miikka.heikkinen@theqtcompany.com>
Plus fixed a QML documentation bug, and added a missing
inheritance to docs.
Task-number: QTRD-3383
Change-Id: I1f7c56c49b829fee2cfd0731bbcc7e248a06a0c0
Reviewed-by: Miikka Heikkinen <miikka.heikkinen@theqtcompany.com>
Change-Id: I742a5a5d7c69ef73ba469bfa4bac49a71b8f578a
Reviewed-by: Mika Salmela <mika.salmela@theqtcompany.com>
Reviewed-by: Tomi Korpipää <tomi.korpipaa@digia.com>
Also added skeleton for remaining QML tests and some
inheritance updates to documentation.
Task-number: QTRD-3368
Change-Id: I28cb52c4738953b7f281ce91a3764f849643d35a
Change-Id: I28cb52c4738953b7f281ce91a3764f849643d35a
Reviewed-by: Miikka Heikkinen <miikka.heikkinen@theqtcompany.com>
If the context exists at renderer deletion, it is possible that
it no longer has valid surface, which means it won't be possible
to set it back to current if we change context for deletion cleanup.
Since the current context will be one of our shared contexts anyway,
there is no need to do a dummy context for cleanup unless the current
context is null.
Change-Id: Ibabe081742beb975ee848ccb3690703ef5b027a9
Reviewed-by: Mika Salmela <mika.salmela@digia.com>
Reviewed-by: Tomi Korpipää <tomi.korpipaa@digia.com>
Bug in adjustment logic caused adjustment to happen pretty much always.
Task-number: QTRD-3374
Change-Id: Id6151d2d3eb8369bbf9252e4eb4b76ed57fb1f3c
Reviewed-by: Mika Salmela <mika.salmela@digia.com>
When static optimization is in use with mesh objects and range
gradient, and the object Y-value resolves into a texture coordinate
that is exactly on the texel boundary, the rendered fragments
of the object are not all same colors on some graphics cards,
despite all vertices having the same UV value.
Fixed by adjusting the Y-value slightly if it is close to the
boundary.
Task-number: QTRD-3370
Change-Id: Ie028602cbd9a00bb0e17049eb8f40feb8b18a6bf
Reviewed-by: Tomi Korpipää <tomi.korpipaa@digia.com>
Context was not getting properly cleaned up in cases where
render thread stopped before the cleanup took place.
Now we ensure that required cleanup is done before context
thread terminates.
Change-Id: I1489914dafec928eebb69bac737f6f858ff49432
Reviewed-by: Tomi Korpipää <tomi.korpipaa@digia.com>
Task-number: QTRD-3368
Just a skeleton with a very basic test set for now.
Change-Id: I53309374a7d9cbfd06a4cdfbeb302f6649db9c4a
Reviewed-by: Miikka Heikkinen <miikka.heikkinen@theqtcompany.com>
Just a skeleton for now.
Change-Id: I27d577062e7e96880d843a87b71791f896bc2522
Change-Id: I27d577062e7e96880d843a87b71791f896bc2522
Reviewed-by: Miikka Heikkinen <miikka.heikkinen@theqtcompany.com>
QtQuick's threaded renderer doesn't seem to always pair up sync and
render (or more accurately, beforeSynchronizing and node's
preprocess calls). Sometimes sync comes without a followup render,
or there are two syncs in a row, even though the sync is using
direct connection. Both of these cases broke the old
logic for handling position based queries.
Changed the logic to actually ensure we have resolved the query
before emitting the relevant signal.
Task-number: QTRD-3358
Change-Id: Ica6c8c311f53a06311c21532aaabc18c28556655
Reviewed-by: Tomi Korpipää <tomi.korpipaa@digia.com>
Miikka Heikkinen
Mika Salmela
Tomi Korpipää
Mika Salmela
Miikka Heikkinen