Take the argument by value instead of rvalue ref. In C++17, this is as
efficient, and enables the
return big;
to use the QByteArray move constructor, avoiding the need for a manual
std::move(), which always looks suspicious on a return statement,
because more often than not, it's a pessimization that breaks NRVO.
This code doesn't seem to exist in Qt 6.2, so only
Pick-to: 6.3
Change-Id: I8bf678102f5df1870cfc61090d12f327478d74d1
Reviewed-by: Fabian Kosmale <fabian.kosmale@qt.io>
Clang 10 warns about unused results of relational operators, which is
where this is coming from.
Fix by adding the usual prefix
[[maybe_unused]] auto r = ~~~;
to silence the warning. Do this elsewhere, too, since [[nodiscard]] is
slowly being rolled out across all our APIs. This is not a complete
sweep, though.
Not picking to 5.15, because this pattern doesn't work there and I
don't want to introduce a new one.
Pick-to: 6.3 6.2
Change-Id: I40dd8ad07496b686979dce533e044cbb486e30f3
Reviewed-by: Fabian Kosmale <fabian.kosmale@qt.io>
The timeToTest constant doesn't need to be captured, claims Clang.
Since I don't recall seeing this warning on GCC, be pragmatic and fix
by letting the compiler choose what to capture: [&]. timeToTest is
const, so it doesn't matter whether we capture by reference or value,
the lambda cannot change it either way.
This code doesn't seem to exist in 5.15, so merely
Pick-to: 6.3 6.2
Change-Id: I48d42ab13ed22ac5eb512dc61235b72a19636ea3
Reviewed-by: Fabian Kosmale <fabian.kosmale@qt.io>
Reviewed-by: Mårten Nordheim <marten.nordheim@qt.io>
The function name suggests that the *= operator is to be benchmarked,
not simple multiplication.
Use the correct operator.
Pick-to: 6.3 6.2 5.15
Change-Id: I6718e8aea640a153083858b39963199e7bab26e9
Reviewed-by: Fabian Kosmale <fabian.kosmale@qt.io>
Reviewed-by: Mårten Nordheim <marten.nordheim@qt.io>
Continuations were using QFutureInterface to create and return the
associated future to the user. Attaching a continuation to the returned
future could cause memory leaks (described in an earlier commit). Use a
QPromise when saving the continuation, to make sure that the attached
continuation is cleaned in the destructor of the associated QPromise, so
that it doesn't keep any ref-counted copies to the shared data, thus
preventing it from being deleted.
Task-number: QTBUG-99534
Pick-to: 6.3 6.2
Change-Id: I52d5501292095d41d1e060b7dd140c8e5d01335c
Reviewed-by: Andrei Golubev <andrei.golubev@qt.io>
Reviewed-by: Mårten Nordheim <marten.nordheim@qt.io>
Reviewed-by: Edward Welbourne <edward.welbourne@qt.io>
It's used by the lancebench and the lance tool, and it will probably be
useful for writing some high-dpi related unit and baseline test cases,
so move it to the shared folder.
Change-Id: I969bab51c9504be13b4c192b4f29f69cd9102868
Reviewed-by: Eirik Aavitsland <eirik.aavitsland@qt.io>
The benchmark was making assumptions about number of
constructor/assignment operator calls, which are not valid in Qt 6,
after the implementation of QList has changed. Considering that we
already check number of constructions, copy constructions, etc., in
tst_qlist.cpp, remove the checks from the benchmark.
As a driveby, fix the following warning:
"warning: parameter 'i' shadows member inherited from type 'MyBase'"
Pick-to: 6.2
Fixes: QTBUG-95096
Change-Id: Ida68fa5803641c8fa84f8309c0093986ed4c0a2b
Reviewed-by: Edward Welbourne <edward.welbourne@qt.io>
Reviewed-by: Andrei Golubev <andrei.golubev@qt.io>
Open the file only if matching on content is needed.
Use QFileInfo::filePath() instead of QFileInfo::absoluteFilePath() in
QMimeDatabase::mimeTypeForFile(). filePath() does much less work, and so
is faster. Thiago Macieira helpfully explained in a review comment why
the absolute path is not useful for correctness here: "Nothing needs
absolute paths within the same application that would resolve the
relative path to absolute. You only need an absolute path if you're
communicating with another application that may be in a different
directory."
QMimeDatabase::mimeTypeForFile() checks fileInfo.isDir(), so the
fileName.endsWith(QLatin1Char('/')) check in
QMimeDatabasePrivate::mimeTypeForFileNameAndData() was redundant when
called from this function. The other two callers of that function now
check this condition before opening IO devices. This improves
performance of the two QMimeDatabase::mimeTypeForFileNameAndData()
overloads in the corner case.
Refactor and optimize QMimeDatabasePrivate::findByFileName() and its
usages. Previously each caller constructed a QFileInfo object and passed
QFileInfo::fileName() into this function. Now the callers simply pass an
absolute or relative path to a file into this function, which then uses
QFileSystemEntry::fileName() to exclude the path. Constructing QFileInfo
is relatively expensive, so this change slightly improves performance.
Optimize QMimeDatabasePrivate::loadProviders() by calling static
QFileInfo::exists() instead of constructing a QFileInfo object and
calling the non-static QFileInfo::exists() overload. Note that the
QFileInfo object was always created, even if QFileInfo::exists() under
an `if` and an `#if` was never called.
The following table contains the average results of the added benchmark
tst_QMimeDatabase::benchMimeTypeForFile() on my GNU/Linux system before
and at this commit. The numbers denote milliseconds per iteration.
data row tag before at
MatchDefault:
archive 0.029 0.016
OpenDocument Text 0.029 0.015
existent archive with extension 0.039 0.025
existent C with extension 0.033 0.020
existent text file with extension 0.033 0.020
existent C w/o extension 0.076 0.074
existent patch w/o extension 0.11 0.105
existent archive w/o extension 0.069 0.066
MatchExtension:
archive 0.012 0.0115
OpenDocument Text 0.0115 0.011
existent archive with extension 0.017 0.016
existent C with extension 0.011 0.011
existent text file with extension 0.011 0.011
existent C w/o extension 0.016 0.0155
existent patch w/o extension 0.013 0.012
existent archive w/o extension 0.013 0.012
MatchContent:
archive 0.019 0.012
OpenDocument Text 0.019 0.012
existent archive with extension 0.053 0.051
existent C with extension 0.056 0.0545
existent text file with extension 0.058 0.056
existent C w/o extension 0.0605 0.059
existent patch w/o extension 0.10 0.099
existent archive w/o extension 0.057 0.054
Change-Id: Idb541656e073a2c4822ace3f4da412f29f2351f8
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
Reviewed-by: David Faure <david.faure@kdab.com>
The new argument allows atomic creation of files with non-default
permissions.
Task-number: QTBUG-79750
Change-Id: I4c49455b41f924ba87148302c8d0f77f5de0832b
Reviewed-by: Mårten Nordheim <marten.nordheim@qt.io>
Results on my machine (only forever results):
********* Start testing of tst_QWaitCondition *********
Config: Using QtTest library 6.3.0, Qt 6.3.0 (x86_64-little_endian-lp64 shared (dynamic) release build; by GCC 11.2.1 20211115), ubuntu 20.04
PASS : tst_QWaitCondition::oscillate_QWaitCondition_QMutex(forever)
RESULT : tst_QWaitCondition::oscillate_QWaitCondition_QMutex():"forever":
637 msecs per iteration (total: 637, iterations: 1)
PASS : tst_QWaitCondition::oscillate_QWaitCondition_QReadWriteLock(forever)
RESULT : tst_QWaitCondition::oscillate_QWaitCondition_QReadWriteLock():"forever":
909 msecs per iteration (total: 909, iterations: 1)
PASS : tst_QWaitCondition::oscillate_std_condition_variable_std_mutex(forever)
RESULT : tst_QWaitCondition::oscillate_std_condition_variable_std_mutex():"forever":
331 msecs per iteration (total: 331, iterations: 1)
PASS : tst_QWaitCondition::oscillate_std_condition_variable_any_QMutex(forever)
RESULT : tst_QWaitCondition::oscillate_std_condition_variable_any_QMutex():"forever":
627 msecs per iteration (total: 627, iterations: 1)
PASS : tst_QWaitCondition::oscillate_std_condition_variable_any_QReadWriteLock(forever)
RESULT : tst_QWaitCondition::oscillate_std_condition_variable_any_QReadWriteLock():"forever":
913 msecs per iteration (total: 913, iterations: 1)
~331 vs. ~630ms. A pretty significant win (2x).
Mårten noticed that on Windows, condition_variable::wait_for(x, 0ms)
will not unlock the mutex, which, however, the program requires, so
use a 1ns timeout instead.
Drive-by fixes:
- add override to run() reimplementations
- fix type of timeout member variable (was int, should be unsigned long)
- fix naming of test functions to distinguish better between QMutex
and std::mutex
Change-Id: Ib92310f15fbd58258b2043504642be5f0b860f39
Reviewed-by: Qt CI Bot <qt_ci_bot@qt-project.org>
Reviewed-by: Lars Knoll <lars.knoll@qt.io>
We have some special handling in qt_windows.h,
use it instead of the original windows.h
Change-Id: I12fa45b09d3f2aad355573dce45861d7d28e1d77
Reviewed-by: Qt CI Bot <qt_ci_bot@qt-project.org>
Reviewed-by: Joerg Bornemann <joerg.bornemann@qt.io>
Reviewed-by: Edward Welbourne <edward.welbourne@qt.io>
In preparation for addition of new baseline tests, establish a new
test category, "baseline". This is similar to the category
"benchmarks" in that it contains tests that use the QTest framework,
but conceptually are not unit tests, in contrast to those under auto/.
Move the existing QPainter baseline test, tst_lancelot, into this new
category, and rename it accordingly.
Baseline tests use the QBaselineTest extension to QTest. Move that
extension too into the tests/baseline directory, allowing the clean
out of the baselineserver directory.
Pick-to: 6.2
Change-Id: I1b527f5867c953b1d22be73798fcf7d1494712ea
Reviewed-by: Qt CI Bot <qt_ci_bot@qt-project.org>
Reviewed-by: Eirik Aavitsland <eirik.aavitsland@qt.io>
Reviewed-by: Volker Hilsheimer <volker.hilsheimer@qt.io>
And bump NTDDI_VERSION to 0x0A00000B (NTDDI_WIN10_CO) at the same time,
to unblock the developers from accessing the latest Windows APIs.
Pick-to: 6.2
Change-Id: Ifbc28c8f8b073866871685c020301f5f20dc9591
Reviewed-by: Qt CI Bot <qt_ci_bot@qt-project.org>
Reviewed-by: Joerg Bornemann <joerg.bornemann@qt.io>
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
Qt requires a compiler that support C++17 thus __cplusplus
is always 201703L or higher. This patch removes checks
for __cplusplus value that always succeed.
Change-Id: I4b830683ecefab8f913d8b09604086d53209d2e3
Reviewed-by: Fabian Kosmale <fabian.kosmale@qt.io>
Reviewed-by: Edward Welbourne <edward.welbourne@qt.io>
Reviewed-by: Qt CI Bot <qt_ci_bot@qt-project.org>
Make file names match CMake's test names (and those follow dir-name)
and class names follow tst_ClassName pattern when testing
ClassName. Purge comments about the qmake configs the CMakeLists.txt
are generated from. Purge empty constructors and init/cleanup methods
of classes. Fix petty coding style violations.
Add qdir/tree/, qurl, qbench and qset benchmarks to their parent directories'
lists of subdirs. Fix unused return error from qurl benchmark.
Change-Id: Ifc15a3a46e71cf82ad0637753517e0df34049763
Reviewed-by: Andrei Golubev <andrei.golubev@qt.io>
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
It tried to find its test-data-directory locally, instead of asking
FileSystem where it was keeping its files, relative to a temporary
directory. Its expected counts didn't match the arcane results of its
data file; since this is a benchmark, not a regression test, I'm
assuming that's just because it was never right or the rules for
QDirIterator's filtering have changed.
Tidied up somewhat in the process.
Change-Id: Ib2dcd314b11cb0f1a6fc425633afd0e9c39bd036
Reviewed-by: Marc Mutz <marc.mutz@kdab.com>
QFont should be constructed with a list of font families.
Pick-to: 6.2
Change-Id: I61141b25d3f6e25f4fea141acbfa8e164d7af58f
Reviewed-by: Andy Shaw <andy.shaw@qt.io>
The QtTest best practices documentations recommends using output
mechanisms such as qDebug() and qWarning() for diagnostic messages,
and this is also what most of our own tests do.
The QWARN() macro and corresponding internal QTest::qWarn() function
was added when QtTest was first implemented, but was likely meant as
an internal implementation detail, like its cousin QTestLog::info(),
which does not have any corresponding macro.
This theory is backed by our own QtTest self-test (tst_silent)
describing the output from QWARN() as "an internal testlib warning".
The only difference between QWARN() and qWarning(), besides the much
richer feature set of the latter, is that qWarning() will not pass
on file and line number information in release mode, but QWARN() will.
This is an acceptable loss of functionality, considering that the user
can override this behavior by defining QT_MESSAGELOGCONTEXT.
[ChangeLog][QtTest] QWARN() has been deprecated in favor of qWarning()
Pick-to: 6.2
Change-Id: I5a2431ce48c47392244560dd520953b9fc735c85
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
The vsnprintf we use in QTest doesn't not have a portable %ls:
It accepts wchar_t, so it's UTF-32 on Linux and UTF-16 on Windows
Change-Id: I9ebda1e92b6e8e4dbbb79c6f2e35a833c587a089
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
C++20 deprecated compound volatile statements such as pre- and
post-increments, to stress that they're not atomic. So instead of
volatile i;
~~~~;
++i;
you're now supposed to write
volatile i;
~~~~;
int j = i; // volatile load
++j;
i = j; // volatile store
which matches more closely what hardware does.
Instead of fixing every use of volatile pre- or post-increment in this
fashion individually, and realising that probably a few more Qt
modules will have the same kind of code patterns in them, write
QtPrivate functions to do the job centrally.
Change-Id: I838097bd484ef2118c071726963f103c080d2ba5
Reviewed-by: Lars Knoll <lars.knoll@qt.io>
Before this change, next() was the only way to advance the iterator,
whether the caller was ultimately interested in just the filePath()
(good) or not (bad luck, had to call .fileInfo()).
Add a new function, nextFileInfo(), with returns fileInfo() instead.
Incidentally, the returned object has already been constructed as part
of advance()ing the iterator, so the new function is faster than
next() even if the result is ignored, because we're not calculating a
QString result the caller may not be interested in.
Use the new function around the code.
Fix a couple of cases of next(); fileInfo().filePath() (just use
next()'s return value) as a drive-by.
[ChangeLog][QtCore][QDirIterator] Added nextFileInfo(), which is like
next(), but returns fileInfo() instead of filePath().
Change-Id: I601220575961169b44139fc55b9eae6c3197afb4
Reviewed-by: Mårten Nordheim <marten.nordheim@qt.io>
In two cases, it was as easy as replacing an unnamed enum's values
with constexpr variables. In the case of QSimplex, I opted for
qToUnderlying(), as the enum made sense on its own.
Change-Id: Ifcf5be14bd2f35e50adabdbd7ecdb2e83f6bf5b4
Reviewed-by: Qt CI Bot <qt_ci_bot@qt-project.org>
Reviewed-by: Mårten Nordheim <marten.nordheim@qt.io>
Reviewed-by: Edward Welbourne <edward.welbourne@qt.io>
There's no point mentioning empty init(), constructor and destructor.
Change-Id: I0b820f62fd46a955aae891adfc68ca366ca60672
Reviewed-by: Andrei Golubev <andrei.golubev@qt.io>
Two QSqlRecord benchmarks that are only relevant for PostgreSQL were
being run for all backends, without producing useful results for the
others. Since the test is data-driven and the generic data-table code
can take a backend-name to decide which to include, pass a suitable
string to the generic data method instead, so that we now simply skip
these tests (and say we're doing so) rather than "passing" them.
Change-Id: I2223c16007a7095a9cadd13a9b2d46813507a35f
Reviewed-by: Andy Shaw <andy.shaw@qt.io>
Because QBENCHMARK re-runs its block repeatedly, to get sensible data,
the block needs to actually do something when repeated. Since these
tests had blocks that looped while (qry.next()), they left qry at its
end state, so such repeats tested nothing. Use seek(0) at the start of
each cycle to actually do the work repeatedly when the block is
repeated. As a drive-by, split a long line.
Pick-to: 6.2 6.1 5.15
Task-number: QTBUG-91713
Change-Id: Id46f77dc5e71335871af79ff61e1980b5f636179
Reviewed-by: Andy Shaw <andy.shaw@qt.io>
At 1000, the set-up was taking longer than the five minutes
QtTestLib's WatchDog allows, so the test got killed.
Pick-to: 6.2 6.1 5.15
Task-number: QTBUG-91713
Change-Id: Ia3c85b223fc917ad5817364505cbffe50d67ddc6
Reviewed-by: Andy Shaw <andy.shaw@qt.io>
Now that this event loop pays attention to test failures, we can avoid
the time-outs that used to happen on test failure. Also check for
premature failures (but don't return early, so we can shut down the
server gracefully) and give the event-loops sensible time-outs.
Task-number: QTBUG-91713
Change-Id: Ib895a5fba0f22654c7fecf996f23649a4b5ce0de
Reviewed-by: Alex Trotsenko <alex1973tr@gmail.com>
Just in case the test isn't testing what we think it is.
One of my earlier changes didn't until this told me about it.
Change-Id: Idd6f415d543509cabb3a64219736bb43e60a70ef
Reviewed-by: Giuseppe D'Angelo <giuseppe.dangelo@kdab.com>
Reviewed-by: Andrei Golubev <andrei.golubev@qt.io>
The sum of the first 100000 naturals is more than 2^32, so using an
int accumulator to collect the values is susceptible to overflow,
which is UB for signed integral types. So switch to an unsigned type.
We don't care about the actual sum, only having the various map
entries we fetch "used".
Since unsigned arithmetic is well-defined even when it overflows, we
can calculate the expected sum and verify it, to ensure that no matter
how clever the optimizer, it won't throw out the accumulator as
written but not read (and then optimize out all the tested code).
As a drive-by, rename one of the accumulators to match the rest.
Change-Id: I93a2825247c96ca88fe52fdb7ce1e5456eebad54
Reviewed-by: Andrei Golubev <andrei.golubev@qt.io>
Reviewed-by: Giuseppe D'Angelo <giuseppe.dangelo@kdab.com>
Mostly so that I can put comments on them to indicate why some tests
are bigger and others not so big.
Change-Id: I633ceb264aa96ee8f5345e3f342a518e8ae4838b
Reviewed-by: Giuseppe D'Angelo <giuseppe.dangelo@kdab.com>
Reviewed-by: Andrei Golubev <andrei.golubev@qt.io>
While we're here:
- remove the GCC precondition for compiling this benchmark.
Task-number: QTBUG-88484
Change-Id: I14f3ea7e4708e274d032a6297e9d4a87ae5dc1c0
Reviewed-by: Edward Welbourne <edward.welbourne@qt.io>
Follow-up to commit 915be6606e, catching
some benchmarks that took for granted they can assign an arbitrary int
to QChar. Since 6.0 this has triggered an assertion.
Given the choice between limiting the range (from 100000 to 0x10000)
and actually handling the out-of-range values as UCS-4 data, the
latter seemed like a more interesting test.
At the same time, take the construction of the strings out of the
loop, as that's not a QMap performance matter, it's a QString one.
Pick-to: 6.1 6.2
Task-number: QTBUG-91713
Change-Id: Id6abab08b5c879f0f764350f66d6aa1dd9f1620a
Reviewed-by: Andrei Golubev <andrei.golubev@qt.io>
Reviewed-by: Giuseppe D'Angelo <giuseppe.dangelo@kdab.com>
Change a += to simple assignment where it's string arithmetic (a
hundred thousand concatenations of "Hello World" add up to more than a
megabyte, in an incremental growth that's going to dominate the QMap
operations we were meant to be benchmarking) and the only reason for
it is to avoid an unused result warning. Accumulating int values is
harmless, but strings are another story !
Pick-to: 6.1 6.2
Task-number: QTBUG-91713
Change-Id: Ib0dc131b0cc75fea23998afc0300e8cb60076c7e
Reviewed-by: Andrei Golubev <andrei.golubev@qt.io>
'main.cpp' is quite generic when using search to look for it.
Change-Id: I547ba16a11db8efb7d4410b94343b03d30da6513
Reviewed-by: Edward Welbourne <edward.welbourne@qt.io>
The usual pattern (at least in corelib) is tst_bench_[lowercased
class-name] for the test and the same with .cpp for the source-file
name. So s/(main|tst_bench_vector)/tst_bench_qvector/g
Change-Id: Ic9bd3ac87adfaec189409c2259cc674ebcec602c
Reviewed-by: Volker Hilsheimer <volker.hilsheimer@qt.io>
Purge comments showing assember from an antique g++ version,
skip #if-ery that's always been on, so makes no difference,
rename two single-letter variables shared between files,
move some extern declarations to a header,
wrap parts of a source file in QT_{BEGIN,END}_NAMESPACE,
add a TODO against an antique commented out #if-ery kludge.
Change-Id: Ic4781960e0c9838027c21d3d392a50f29598132c
Reviewed-by: Volker Hilsheimer <volker.hilsheimer@qt.io>
The test constructs an ersatz object with which to overwrite a
QVector's internals, but based it on Qt 5's memory layout. Since Qt 6
completely rearranged that memory layout, the test now crashes (or, in
a debug build, trips over an out-of-bounds assertion).
So suppress the test until those who perpetrated it and/or the
reworking of QVector's internals can work out how to fix it.
That way, QA can at least run the benchmark without crashing.
Task-number: QTBUG-95061
Pick-to: 6.2
Change-Id: I0e8c8d58f2002497f6e29d0ad25f840e207704a4
Reviewed-by: Volker Hilsheimer <volker.hilsheimer@qt.io>
Let's have fewer files with the same main.cpp name.
Change-Id: Iad94478affcca73a25ed29ba4380f1b3ca4f329d
Reviewed-by: Volker Hilsheimer <volker.hilsheimer@qt.io>
