It wasn't tested and behaves in a very particular way.
Change-Id: I60a31681e5b221cf9a86df77e410a76ee4c10864
Reviewed-by: Eirik Aavitsland <eirik.aavitsland@qt.io>
This is not to be taken as supported and is still undefined behavior,
but I prefer we do not crash.
Change-Id: Icf4f3398bfd57fcbdc611a5a821a1f2de0838330
Reviewed-by: Eirik Aavitsland <eirik.aavitsland@qt.io>
The previous implementation was *extremely* expensive. It
relied on loading a binary JSON file from resources (which
involved decompressing it), then extracting information out of
it to build a gradient. Already-loaded gradients were kept in
a local cache, which had to be mutex protected.
Instead, this patch extends the gradient generator to build
static arrays filled with the web gradient data, sitting in
.rodata.
These arrays are used when building QGradient objects with a
web gradient. No explicit mutex protection is necessary, since
accesses will just read from the arrays.
As benefits, this patch removes:
* the binary json representation from QtGui's resources (~4KB
compressed, ~50KB uncompressed)
* the overhead of reading from the JSON for each used web
gradient;
* the startup costs of registering the webgradients in the
resources;
* all the overhead of mutex locking when building such
gradients;
* all the runtime memory allocations to load, parse and cache
the web gradients (including the memory + CPU spike on first
load due to the uncompression of the JSON data, as well as a
couple of deep copies).
Change-Id: If5c3d704430df76ce8faf55ee75ebd4639ba09c4
Reviewed-by: Tor Arne Vestbø <tor.arne.vestbo@qt.io>
Reviewed-by: Ulf Hermann <ulf.hermann@qt.io>
Reviewed-by: Qt CI Bot <qt_ci_bot@qt-project.org>
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
Reviewed-by: Edward Welbourne <edward.welbourne@qt.io>
When an existing item is replaced with a new one in
QStandardItemModel::setItem() then the persitent index is invalidated
which leads to some unexpected behaviors (like e.g the header size and
resize mode are reset).
Therefore we have to make sure that the invalidation does not happen.
This can be achieved by delaying the call to QStandardItem::setModel()
for the old item until the new is properly added. After this, the old
item no longer gets a valid QModelIndex from the model and therefore
can't invalidate the persistent index anymore.
Fixes: QTBUG-13605
Fixes: QTBUG-73000
Fixes: QTBUG-80586
Change-Id: I4e45e6feb81b7287c0859f638d7ab1a576fc2f0f
Reviewed-by: David Faure <david.faure@kdab.com>
When cloning a QTextDocument, the text fragment of the original document
is copied into the new one, which results into copying also the
formatting attributes. However, when the text document is empty, the
corresponding text fragment is also empty, so nothing is copied.
If we want to transfer the formatting attributes for an empty document,
we need to set them explicitly.
Fixes: QTBUG-80399
Change-Id: I382cd0821723436120af47c06ec7bfa849636307
Reviewed-by: Cristian Maureira-Fredes <cristian.maureira-fredes@qt.io>
Reviewed-by: Eirik Aavitsland <eirik.aavitsland@qt.io>
QDoubleValidator would accept "1,23" as valid in a locale which has ','
as a thousand separator. However, it should have been Intermediate
instead, as there is still one digit missing.
Fixes: QTBUG-75110
Change-Id: I6de90f0b6f1eae95dc8dfc8e5f9658e482e46db3
Reviewed-by: Ulf Hermann <ulf.hermann@qt.io>
Reviewed-by: Edward Welbourne <edward.welbourne@qt.io>
The new version takes/returns a value that can be unpacked and passed to
other functions without knowing which backend is in use.
The old API will be removed in a later change when dependent modules have
been updated
Task-number: QTBUG-78570
Change-Id: I18d928ceef3cb617c0c509ecccb345551a7990af
Reviewed-by: Laszlo Agocs <laszlo.agocs@qt.io>
The code was assuming that if the parsing of the value worked,
then it must be a list of 4 variants. But in this case it's just
a single length.
This came from <td> using 4 values for border-width
while other elements use a single value. But the storage
is shared. So the fix is to use 4 values everywhere.
When reading 4 and there's only one, it gets duplicated,
so the caller can just use the first one in that case.
Task-number: QTBUG-80496
Change-Id: I682244b6e3781c4d673a62d5e6511dac263c58e8
Reviewed-by: Nils Jeisecke <nils.jeisecke@saltation.com>
Reviewed-by: Shawn Rutledge <shawn.rutledge@qt.io>
The code was initially introduced in
9204b8c31e but getting the names were
conditioned on whether or not Windows identified it as a truetype font.
This excluded cases which had preferred names embedded but was not
truetype fonts. To fix that we run the code unconditionally.
[ChangeLog][Windows] Fixed a bug where some fonts would not be
accessible by referencing their typographic name.
Fixes: QTBUG-78556
Change-Id: I8823684b09cce3b1b8722b1e609a5bb49b13da13
Reviewed-by: Eskil Abrahamsen Blomfeldt <eskil.abrahamsen-blomfeldt@qt.io>
Reviewed-by: Allan Sandfeld Jensen <allan.jensen@qt.io>
For Metal and Vulkan this needs actual work because that's where
the concept of renderpass descriptors is relevant. GL and D3D can
just return true always.
The big benefit of this is that Qt Quick can now compare renderpass
descriptors via isCompatible() for its pipeline cache (similarly to
how it is already using isLayoutCompatible() for srbs), and so
renderpass descriptors for layers (Item.layer, ShaderEffect) will
typically be compatible and so can pick up pipelines created by other
layers from the cache.
Also add autotests for shader resource binding and renderpass descriptor
compatibility.
Task-number: QTBUG-80318
Change-Id: I0008bc51c4ee13b0113d2c8caf799e1257f18a18
Reviewed-by: Paul Olav Tvete <paul.tvete@qt.io>
While we are at it, remove the Border and MirrorOnce wrap modes that have
not been supported on OpenGL, because they are unsupported with Metal+iOS
as well.
Task-number: QTBUG-78580
Change-Id: I0db94b9d3a6125b3bb5d7b1db5d02a42cd94d2c2
Reviewed-by: Tor Arne Vestbø <tor.arne.vestbo@qt.io>
Reimplement QDomDocument using QXmlStreamReader and switch to the new
implementation starting from Qt 6.
The changes in the behavior are reflected in tests: some test cases
which were marked as "expected to fail" are now passing.
Task-number: QTBUG-76178
Change-Id: I5ace2f13c036a9a778de922b47a1ce35957ce5f6
Reviewed-by: Kai Koehne <kai.koehne@qt.io>
This was used to support QFlags f = 0 initialization, but with 0 used
as a pointer literal now considered bad form, it had been changed many
places to QFlags f = nullptr, which is meaningless and confusing.
Change-Id: I4bc592151c255dc5cab1a232615caecc520f02e8
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
It was converted over ARGB32PM, when it should have been directly
converted to not lose accuracy, instead there was an unnecessary direct
ARGB32->RGB30 conversion, which was converted to the necessary type.
This also improves the selection of conversion over ARGB32PM or RGBA64PM
for ARGB32 and RGBA8888 by using 32-bit conversion when alpha is not
relevant.
Change-Id: I5990d8a23b2909d3910d8c1213fa46477742b052
Reviewed-by: Eirik Aavitsland <eirik.aavitsland@qt.io>
You can save a "skeletal" document with list items to fill in later,
the same as you can do in HTML or ODF format. Reading them back via
QTextDocument::fromMarkdown() isn't always perfect though.
Fixes: QTBUG-79217
Change-Id: Iacdb3e6792250ebdead05f314c9e3d00546eeb9f
Reviewed-by: Shawn Rutledge <shawn.rutledge@qt.io>
The markdown parser generates empty lists in some cases when a character
that can be used as a bullet is found on a line by itself.
cbEnterBlock() and cbLeaveBlock() are called symmetrically in such cases.
QStack::pop() on an empty stack triggers an assert, so push and pop need
to be done symmetrically too. But it's difficult to actually create the
list as soon as the MD_BLOCK_UL or MD_BLOCK_OL callback occurs, without
breaking the case fixed in 7224d0e427 (and
probably other cases). That's because QTextCursor::insertList() creates
a list item at the same time as it creates the list itself, and also
inherits block formatting from the previous block. We now insert empty
lists with empty items whenever the need for that is detected though,
and there's a failsafe to prevent popping in case something still goes
wrong with that logic. We aren't strict about reproducing the original
markdown when regenerating it via toMarkdown(), but it's getting closer.
Fixes: QTBUG-78870
Change-Id: Ided194ce7aec2710c60dbac42761ee4169ed9b78
Reviewed-by: Edward Welbourne <edward.welbourne@qt.io>
Reviewed-by: Robert Loehning <robert.loehning@qt.io>
If a font with only a family set is resolved with one that has been setup
with setFamilies() then the family needs to be prepended to the families
list after resolving. This is so that the font still prefers the one set
as just a family with no famillies set.
This also amends the QFontDialog test to account for this too.
[ChangeLog][QtGui][Text] Resolving a font that just has a family set
with families set will prepend the family to the families so that it
is still the first preference for the font.
Task-number: QTBUG-46322
Change-Id: Icc4005732f95b2b4c684e592b06b31e133270e44
Reviewed-by: Eskil Abrahamsen Blomfeldt <eskil.abrahamsen-blomfeldt@qt.io>
In qmake syntax there should be a colon or curly braces following a condition.
Change-Id: Ibd989662aef6320cec8093e7c6103bf8362b8255
Reviewed-by: Liang Qi <liang.qi@qt.io>
Call adjustedFormat() as advised by the docs: "Applications are advised
to set this format on their QWindow in order to avoid potential BAD_MATCH
failures."
Task-number: QTBUG-79659
Change-Id: Ibf415fb0ee64bdd3f01d4ba744244bce811c0d27
Reviewed-by: Paul Olav Tvete <paul.tvete@qt.io>
Binary JSON is said to become deprecated. Therefore, add support
for CBOR. Binary JSON is still supported for deserialization, so
all existing .qsb files will continue to work, as long as the
binaryjson feature is enabled in the Qt build.
Also makes QShaderDescription comparable. This is important for
tests in particular.
A nice side effect of using CBOR is that .qsb files become smaller.
For a typical Qt Quick material shader this can mean a reduction of
300 bytes or more.
Task-number: QTBUG-79576
Change-Id: I5547c0266e3e8128c9653e954e47487352267f71
Reviewed-by: Paul Olav Tvete <paul.tvete@qt.io>
During the creation of a raster paint engine in QImage::paintEngine(),
the QImage will be detached. At least old gcc versions would get
confused so that the newly created paintengine would end up in the old
QImage copy insteads of the newly detached one. Work around by
dropping the temporary engine pointer.
Fixes: QTBUG-79383
Change-Id: I27b1f24312269bc2bcc641dc4334397a92e3bfbb
Reviewed-by: Allan Sandfeld Jensen <allan.jensen@qt.io>
The default font on Windows ("Times" is not found) does not have
a ligature for "fi", so the test would not actually be testing
what it was supposed to on this platform, and would pass even
when the code was buggy.
To enable the test on Windows, we select a standard font which
has the ligature (Calibri).
Change-Id: Ic117cd8e549aa729a0cd68006d7c180c6c89c053
Reviewed-by: Simon Hausmann <simon.hausmann@qt.io>
For some scalings, setClipRect(QRect) would produce a clip one pixel
different from setClipRect(QRectF) because of different
rounding. Ditto for setClipRegion. Fix by making sure to transform
QRectFs instead of QRects.
Fixes: QTBUG-78962
Fixes: QTBUG-78963
Change-Id: I0be721133858c30769ec6d81e978962a3d6b70cf
Reviewed-by: Christoph Cullmann <cullmann@kde.org>
Reviewed-by: Allan Sandfeld Jensen <allan.jensen@qt.io>
CMake will complain that we have a duplicate test name otherwise.
Generally it makes a lot of sense to name the test binary the same as
the test itself.
Change-Id: I27c4b51e6a2869f025e1100f1a9dd6b54ebdaf55
Reviewed-by: Leander Beernaert <leander.beernaert@qt.io>
Reviewed-by: Mitch Curtis <mitch.curtis@qt.io>
Reviewed-by: Morten Kristensen <me@mortens.dev>
Reviewed-by: Liang Qi <liang.qi@qt.io>
QDesktopWidget is marked as obsolete in docs, but it is not yet
completely deprecated, some of its methods are still in use.
Replace uses of the following methods marked as obsolete:
- QDesktopWidget::screenNumber(QWidget*) -> QWidget::screen()
- QDesktopWidget::screenGeometry(QWidget*) -> QWidget::screen()->geometry()
- QDesktopWidget::availableGeometry(QWidget*) -> QWidget::screen()->availableGeometry()
Task-number: QTBUG-76491
Change-Id: I2cca30f2b4caa6e6848e8190e09f959d2c272f33
Reviewed-by: Tor Arne Vestbø <tor.arne.vestbo@qt.io>
There is no onscreen support for WinRT in the D3D11 backend yet.
However, offscreen operations (rendering into a texture) should work.
One catch is that there is no D3DCompile available for deployed WinRT
apps. So ship the intermediate format (DXBC output from fxc) in the
.qsb files.
Change-Id: Ic0aba4b817c27d13dcf3af41bf7612d799382655
Reviewed-by: Qt CI Bot <qt_ci_bot@qt-project.org>
Reviewed-by: Laszlo Agocs <laszlo.agocs@qt.io>
Also improve (docs and runtime checks) and test the minimum set
of required data to create a graphics pipeline.
Task-number: QTBUG-78971
Change-Id: If5c14f1ab1ff3cf70f168fde585f05fc9d28ec91
Reviewed-by: Paul Olav Tvete <paul.tvete@qt.io>
The joys of "level - Specifies the mipmap level of the texture
image to be attached, which must be 0." for glFramebufferTexture2D
in OpenGL ES 2.0.
Change-Id: Iaf19502f48d7ba73b26abb72535bfa6696a1e182
Reviewed-by: Paul Olav Tvete <paul.tvete@qt.io>
...and make the Null backend able to deal with these, for RGBA8 textures
at least. Naturally it is all QImage and QPainter under the hood.
Also fix a bug in the OpenGL backend, as discovered by the autotest:
the size from the readback did not reflect the mip level.
Task-number: QTBUG-78971
Change-Id: Ie424b268bf5feb09021099b67068f4418a9b583e
Reviewed-by: Paul Olav Tvete <paul.tvete@qt.io>
This also marks the beginnings of significantly extending autotesting
of the resource and rendering functionality in QRhi.
Also involves fixing up the buffer operation lists like we did
for textures before. This is to ensure updates and reads on the
same batch execute in the correct order. So just have two lists:
one with buffer, one with texture operations.
Also simplify the struct layouts. No need for those inner structs
with many duplicate members. This reduces the size even, since using a
union was never an option here. Also switch to a VLA, the size is around
253 KB per batch.
The Null backend now keeps track of the QRhiBuffer data so it can return
valid results in readbacks.
Task-number: QTBUG-78984
Task-number: QTBUG-78986
Task-number: QTBUG-78971
Task-number: QTBUG-78883
Change-Id: I9694bd7fec523a4e71cf8a5c77c828123ebbb3bd
Reviewed-by: Paul Olav Tvete <paul.tvete@qt.io>
Properly return an invalid frame when calling jumpToFrame()
with a non existent frame number.
Fixes: QTBUG-79029
Change-Id: Ic40f4a6de3106fab42c0bb6c961194be47b04e31
Reviewed-by: Eirik Aavitsland <eirik.aavitsland@qt.io>
