Cost of a type lookup for core built-in types is really small, just few
cpu instructions, but the benchmark was testing create() and destroy()
functions (in a different fashion) which by definition allocate and
de-allocate memory. These memory operations are significantly more
expensive which obfuscate the results.
Change-Id: I33c679f57e6c2b57e98328f076dfe249ab7bcde8
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@digia.com>
Reviewed-by: Stephen Kelly <steveire@gmail.com>
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
This does mostly the same HB-old did, except of non-BMP codepoints
support and better handling of space and punctuation characters.
Change-Id: Iaf8565a35f27d5f4837128cec345eff3d9d2bdf0
Reviewed-by: Lars Knoll <lars.knoll@digia.com>
Reviewed-by: Eskil Abrahamsen Blomfeldt <eskil.abrahamsen-blomfeldt@digia.com>
Factor out function to determine the font family for a style
hint and use that for the FreeType font database as well.
Task-number: QTBUG-39961
Change-Id: Ic5ec0928c7994b61e4d2b10d829826658b1aa1d5
Reviewed-by: Björn Breitmeyer <bjoern.breitmeyer@kdab.com>
Reviewed-by: Friedemann Kleint <Friedemann.Kleint@digia.com>
This is much more useful than the URL "file:", it allows to use
"empty path" and "empty URL" for the same meaning (e.g. not set).
QFileDialog actually uses "file:" though, as the URL for the
"My Computer" item in the sidebar. This patch preserves that.
[ChangeLog][QtCore][QUrl] QUrl::fromLocalFile now returns an empty URL
if the input string is empty.
Change-Id: Ib5ce1a3cdf5f229368e5bcd83c62c1d1ac9f8a17
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
Saves us from an extra load in the loop when the result is already
there.
Change-Id: Ib65b5b13f27975b8fc88be9b888c19505bf5e07c
Reviewed-by: Jędrzej Nowacki <jedrzej.nowacki@digia.com>
Reviewed-by: Olivier Goffart <ogoffart@woboq.com>
This was a long-time coming.
One innovation from this commit is that it will add the source to
SOURCES if the compiler is already generating code for that specific
target. That is currently always the case for Neon, and the MIPS DSPs
since that is the only condition in which configure will enable those
targets. And because of qt_module.prf, it's also always the case for
SSE2 (but not for SSE3 or higher).
So simplify the .pri files by removing always-true conditions.
Change-Id: Ib24af74717b652c9a6be246e3c17a839470f37da
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@digia.com>
Reviewed-by: Erik Verbruggen <erik.verbruggen@digia.com>
We don't actually detect whether the compiler can create Neon code or
provides Neon intrinsics. Most of them do, so that test would be mostly
moot. We removed the detection previously because we couldn't
automatically enable Neon due to leakage of instructions outside the
areas protected at runtime.
Instead, we rely on the mkspec properly passing the necessary flags that
enable Neon support.
This commit does not change that. All it does is verify whether the arch
detection found "neon" as part of the target CPU features. In other
words, it moves the test that was in simd.prf to configure.
It does fix the Neon detection in configure.exe, which was always
failing for trying to run a test that didn't exist
(config.tests/unix/neon).
Change-Id: Id561dfb2db7d3dca7b8c29afef63181693bdc0aa
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@digia.com>
No need to duplicate the extra compiler code.
Change-Id: Ic5656d2f4500c19e0428a7eec1cbcef6c353e99b
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@digia.com>
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
The "rbit" instruction requires ARMv6T2 or higher. This was found in the
CI when building the imx6 target:
Compiler: arm-poky-linux-gnueabi-g++
Flags: -mfloat-abi=hard -mfpu=neon
Errors from the assembler:
{standard input}:3078: Error: selected processor does not support ARM mode `rbit r3,r3'
{standard input}:7341: Error: selected processor does not support ARM mode `rbit ip,ip'
That compiler defaults to ARMv5T. That's obviously wrong for an i.MX 6,
which is a Cortex-A9 (ARMv7), but the correction applies for older
processors.
Change-Id: I56c276fa411977dd7cd867d62adf021e4909302c
Reviewed-by: Jędrzej Nowacki <jedrzej.nowacki@digia.com>
QVariant::convert accepts int as an argument, there was no need to cast.
Change-Id: I774c9567972860d887e17acb91ec332ffcebd9d5
Reviewed-by: Simon Hausmann <simon.hausmann@digia.com>
QJsonValue, while comparing two QJsonArrays, should consult also length
of the arrays, because a different than null base pointer doesn't mean
that an array is not empty.
Change-Id: If76739355a4e74b842e836289565f98d95c006d5
Reviewed-by: Lars Knoll <lars.knoll@digia.com>
QJsonValue, while comparing two QJsonObjects, should consult also length
of the objects, because a different than null base pointer doesn't mean
that an object is not empty.
Change-Id: Ibee1849ef9fed15d32f2c8f2aad9b053846e46b7
Reviewed-by: Lars Knoll <lars.knoll@digia.com>
Before this change such code:
QJsonObject o;
o["blah"];
would create property "blah" and assign null value to it, while
this code:
const QJsonObject o;
o["blah"];
would not. The change unifies the confusing behavior. Now reading
a non-existing property, is not causing a property to be added
in any visible way.
Internally QJsonObject stores a special hash of undefined, but
referenced values. Such reference is supposed to not live long,
only to the first compacting or assignment.
Change-Id: Ib022acf74ff49bad88d45d65d7093c4281d468f1
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
Reviewed-by: Lars Knoll <lars.knoll@digia.com>
The operator should always return an undefined values for an empty
object
Change-Id: Ic38f7660d77c64b2d001967bc5109df4185db74a
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
Reviewed-by: Lars Knoll <lars.knoll@digia.com>
It is a minor reduction, in release build it is ~200 bytes
Change-Id: I4f7972c95769f2e0ca1ddc935ff7a0a6b4379e2a
Reviewed-by: Marc Mutz <marc.mutz@kdab.com>
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@digia.com>
There is no need for a custom algorithm as we can use
std::count
Change-Id: Id1ab514c7cd8f52efe49b27712121415d7ca4455
Reviewed-by: Jędrzej Nowacki <jedrzej.nowacki@digia.com>
Reviewed-by: Marc Mutz <marc.mutz@kdab.com>
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@digia.com>
Fix tests widgets leaking widgets and ensure it stays
that way by adding a check to cleanup().
Task-number: QTBUG-38858
Change-Id: I77a81d823c68cf0b4e51c2da55b1c473d6e4170b
Reviewed-by: Oliver Wolff <oliver.wolff@digia.com>
No actual reason for this test, except my curiority. Then again, it's
good to have this check, too.
Change-Id: I815fce7e4dbe76e21cac29beb1dbfc1083191d24
Reviewed-by: Olivier Goffart <ogoffart@woboq.com>
No actual reason for this test, except my curiority. Then again, it's
good to have this check, too.
Also checks that the last entry in the init_list "wins", which is
not how std:: containers work.
Change-Id: I4f7d1228f2b90a904b6c3f99e54afcd9970b723e
Reviewed-by: Olivier Goffart <ogoffart@woboq.com>
No actual reason for this test, except my curiority. Then again, it's
good to have this check, too.
Also checks that the last entry in the init_list "wins", which is
not how std:: containers work.
Change-Id: Ia284d093cd0029432372630e81657fb687b9516f
Reviewed-by: Olivier Goffart <ogoffart@woboq.com>
This fixes high CPU load for upload devices that don't generate
a constant stream of data. Their readData() function was called all the
time without returning actual data.
This was noticed when implementing an upload device that emits data in
a limited way for bandwidth limiting.
[ChangeLog][QtNetwork][QNetworkAccessManager] Fixed high CPU load when handling
POST/upload QIODevice that generates data on readyRead().
Change-Id: Iefbcb1a21d8aedef1eb11761232dd16a049018dc
Reviewed-by: Richard J. Moore <rich@kde.org>
Although it may seem strange that such a method is const,
optimizing doesn't affect the user-visible part of the object.
Moreover, *not* having it const makes it asymmetrical with other
methods (such as match()) which are const, and under certain
conditions optimize as well.
Change-Id: I0cd8d4a6909d00629fcc65c1c3a1f011f31db782
Reviewed-by: Olivier Goffart <ogoffart@woboq.com>
Better late than never.
Change-Id: If3bbeb4dfe3a8d49ceb02d9c2d0f2eff71595105
Reviewed-by: Leena Miettinen <riitta-leena.miettinen@digia.com>
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@digia.com>
Blocking DBus calls have the potential to totally wreck user interactivity at
best (actions taking too long) and make things appear completely broken at worst
(the default timeout is 30 seconds, which is a huge amount of time, especially
if you get unfortunate and have a repeated blocking call).
Provide a warning when a call is found that takes too long, based on some preset
durations (200ms for the main thread, 500ms for other threads on the basis that
the main thread is generally more important).
Also provide configuration knobs for these environment variables, in miliseconds:
setting them to 0 will warn on all blocking DBus calls. Setting them to -1 (the
default, on release builds) will disable the warning.
[ChangeLog][QtDBus] Blocking calls that take a long time will now generate a
warning. The time taken may be tuned using the environment variables
Q_DBUS_BLOCKING_CALL_MAIN_THREAD_WARNING_MS and
Q_DBUS_BLOCKING_CALL_OTHER_THREAD_WARNING_MS. The value represents (in
milliseconds) how long before a blocking call is warned on. A value below zero
disables the warning, a value of zero will warn on all blocking calls.
Change-Id: I0ab4c34aa01670a154d794d9f2694b3235e789db
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
QString::toUtf8 already does it. I guess I forgot to update this part in
d51130cc3a.
Change-Id: I83feafcb0383758f7e64d5142f57a7ae6a2ff351
Reviewed-by: Olivier Goffart <ogoffart@woboq.com>
[ChangeLog][QtWidgets][QLineEdit] The signal textEdited() is now
emitted when the user clicks the clear button created by
setClearButtonEnabled() as well.
Task-number: QTBUG-40287
Change-Id: Iacd303ffd1533f27cfa68a6120cdd370e3d31ddc
Reviewed-by: Marc Mutz <marc.mutz@kdab.com>
The old unix locking code is no longer working on unix since it locks on
a file descriptor, but QSaveFile creates a new file, and as a result we
get the lock on the wrong file. Also there is no need to keep the lock
held only for reading as QSaveFile is atomic. It just needs to be held
when doing a read before writing.
As a result, since we don't hold the same lock, there could be a race
with an application running an older version of Qt if they are writing
on the same configuration file.
[ChangeLog][QtCore][QSettings] Fixed data loss while writing the config
to the disk fails.
[ChangeLog][Important behavior changes] The locking mechanism inside
QSettings has changed and is no longer compatible with the one of
previous versions of Qt. There might be corruption if two applications
running different versions of Qt are writing to the same config file
at the same time. You must also now have write permissions in the
directory containing the settings file in order to write settings
Task-number: QTBUG-21739
Change-Id: I0844a5e96c8bc1e1222a3dac6cc48170ca77fe1b
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
This file is already listed in QTOBJS.
Change-Id: I5c77620dc5e441435bd31a87a08178e0c82e9c05
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@digia.com>
QRingBuffer is a fully inlined class used in many I/O classes.
So, it must be as fast and small as possible. To this end, a lot of
unnecessary special cases were replaced by generic structures.
Change-Id: Ic189ced3b200924da158ce511d69d324337d01b6
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@digia.com>
This reverts the extractTextAppearanceInformations() changes done in
f7ac2b4 and introduces another method extractTextAppearance() for
extracting a specific TextAppearance.
The problem was that extractTextAppearanceInformations() was overriding
the TextAppearance attributes with TextView attributes. This resulted
in wrong text colors for Switch & CalendarView:
- Switch_switchTextAppearance
- CalendarView_dateTextAppearance
- CalendarView_weekDayTextAppearance
Change-Id: Ib300b6eb14649319b2f07012803a4d031a717b50
Reviewed-by: BogDan Vatra <bogdan@kde.org>
Not just filesSelected+fileSelected (which only happens for local files).
Change-Id: Ife592c3c921231356f96cbc2871b6d724a15d2c8
Reviewed-by: Friedemann Kleint <Friedemann.Kleint@digia.com>
Use the placeholder top level geometry to determine the
screen.
Task-number: QTBUG-40155
Change-Id: If56b239200e9daf754b0963574367cf7891c5f40
Reviewed-by: Giuseppe D'Angelo <giuseppe.dangelo@kdab.com>
This also coincidentally fixes logging of prepared queries, which previously
weren't logged.
Change-Id: I41b3559080662284699ac3dfa4fa4236342c61d2
Done-by: John Brooks <john.brooks@jollamobile.com>
Reviewed-by: Kai Koehne <kai.koehne@digia.com>
The default in QTextStream is to use the C locale, so default
QTextStream are not affected. When you set a QLocale on it, the default
was to use group separators (which is the QLocale default too). This
commit makes QTextStream respect a QLocale in which the
OmitGroupSeparators option had been set.
Task-number: QTBUG-39956
Change-Id: I00fbe12fca7f0287c7217deb487ded6582a03b52
Reviewed-by: Lars Knoll <lars.knoll@digia.com>
Instead of playing clever tricks with the index variable, simply
scan through the list of argument-carrying signals and, if nothing
fits there, fall explicitly back to accepted().
Change-Id: I1be49ce601edb854cc1bab5be994b2c09fe31b3c
Reviewed-by: Friedemann Kleint <Friedemann.Kleint@digia.com>
Otherwise, when interrupting the test half-way (e.g. in gdb),
the user's ~/.config/QtProject.conf would be modified.
Change-Id: Id582c4a3fbf0acbed460126fccf0d39b3a09758a
Reviewed-by: Friedemann Kleint <Friedemann.Kleint@digia.com>
This allows to use the "test mode" of QStandardPaths in unittests,
to stay away from the user's real settings.
Change-Id: I1cb1f63a4bff35dfe236924c4dcd7cf761ee50c1
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@digia.com>
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
