Android 15 (at least) seems to have changed how float arguments are
passed to native functions, breaking our (conceptually correct) code for
processing the va_arg list into a list of static argument types of the
function implementation.
To fix this, we have to move away from using a va_arg function, and
register a function with statically typed arguments instead.
Use a template function that we instantiate with variadic arguments
deduced from the actual function, using a factory-helper that generates
a JNINativeMethod struct with that template instantiation as the
function pointer. Move all of that into a struct where we can also
declare the signature string as compile-time constant without cluttering
the namespace with static objects.
We can now remove the helpers that took care of type promotion in va_arg
functions, and of the tuple-construction from a va_list.
As a drive-by, don't cast function pointers to void *; it's strictly
speaking undefined behavior in C and should have generated a compiler
warning, if not a hard error [1]. We must initialize the
JNINativeMethod::fnPtr member with the address of the function pointer
instead.
[1] https://port70.net/~nsz/c/c11/n1570.html#6.3.2.3p8
Also, declare the native method as the JNICALL calling convention. That
is only defined on Windows, so makes no difference in practice, but it's
the correct thing to do anyway.
Fixes: QTBUG-132410
Change-Id: I190b95fcbcd07cf99c6765fa426c3c351f91994a
Reviewed-by: Volker Krause <vkrause@kde.org>
(cherry picked from commit e91a17873ee4ae58d369b8eb70029cf895b31d03)
Reviewed-by: Qt Cherry-pick Bot <cherrypick_bot@qt-project.org>
(cherry picked from commit 103aa5afb2d95f883daea41715c97455550285b2)
Demonstrates the problem described in the bug report on Android
15/x86_64. QEXPECT_FAIL the relevant test cases until the next
commit fixes the problem.
Task-number: QTBUG-132410
Change-Id: I065fd29282ef42ed75a2ed8177ded183c92aa6e3
Reviewed-by: Volker Hilsheimer <volker.hilsheimer@qt.io>
(cherry picked from commit 749367da8c3309c98b3285836c2bd8abcd7274b1)
Reviewed-by: Qt Cherry-pick Bot <cherrypick_bot@qt-project.org>
(cherry picked from commit e342a5261332c2ef2c3b0d4d14f22dbdbd4a99c3)
It's kinda is important to verify that commit() worked; all the
follow-up tests assume it, so wrap the calls in QVERIFY().
Amends 7e5e7eeaa1.
Pick-to: 6.5 5.15
Change-Id: I01185d65a5ec06ed122e59c43d9cdcc3a7157259
Reviewed-by: Volker Hilsheimer <volker.hilsheimer@qt.io>
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
(cherry picked from commit 97fe38ff0d1e60f92c0bfa6fb3571a0433e60352)
Reviewed-by: Qt Cherry-pick Bot <cherrypick_bot@qt-project.org>
(cherry picked from commit 3660ff9ff88a69ce9f1366e5870f6aa0323f1fda)
To ensure we don't overwrite it in this quite long function.
Amends 7e5e7eeaa1.
Pick-to: 6.5 5.15
Change-Id: Ie5f81b50d2d66ba9e064810f8cbc1a7af298cc2b
Reviewed-by: Volker Hilsheimer <volker.hilsheimer@qt.io>
(cherry picked from commit 67a906b5564a1ccf03e547aa78fc6fd910d9cd15)
Reviewed-by: Qt Cherry-pick Bot <cherrypick_bot@qt-project.org>
(cherry picked from commit b2dad9724d5c92582851e56ba73383644ea8b32b)
The docs said only the first placeholder would be considered; in fact,
it considers only the last.
Fix the docs and add a test (also for QTemporaryDir).
Amends b4bb4449aea7592afdb9b9134bb90c40fe29735a.
While 6.6 and earlier are unaffected by the incorrect documentation,
pick this to all active branches in order to verify the behavior is
the same everywhere.
Pick-to: 6.5 6.2 5.15
Fixes: QTBUG-132597
Change-Id: Ia7a37f714f834191b07420d31ca9024702b537cc
Reviewed-by: Ivan Solovev <ivan.solovev@qt.io>
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
(cherry picked from commit e7abbc7bf3ab06359b2847b1e1a46c38d8bc6581)
Reviewed-by: Qt Cherry-pick Bot <cherrypick_bot@qt-project.org>
(cherry picked from commit 58221c19559541d7da24fa68f05310aeee230e26)
Calling reset() with the owned handle should maintain ownership and not
leak resources. Move assignment should also handle self-assignment
without leaks. The new tests verifies that this holds true.
Task-number: QTBUG-132507
Change-Id: Icdfa4d67b06b71a74e810593e57449b51982c98c
Reviewed-by: Tim Blechmann <tim.blechmann@qt.io>
(cherry picked from commit 03bd9491491881529ad28cd6d672edfdda9a0065)
Reviewed-by: Qt Cherry-pick Bot <cherrypick_bot@qt-project.org>
(cherry picked from commit 33f37ff6376835dc8add4e5d8273f3dc5e7b086d)
Handle traits should be simple, and not require exceptions. This patch
will allow us to improve the noexcept correctness of QUniqueHandle.
Change-Id: I84d92818a2fcea5b98e09c0b7dc08b251751396c
Reviewed-by: Tim Blechmann <tim.blechmann@qt.io>
(cherry picked from commit 3c57c7357422bdfda60f56901a7590b258915b6b)
Reviewed-by: Qt Cherry-pick Bot <cherrypick_bot@qt-project.org>
(cherry picked from commit 8adfacca6184d439c5febd71a6c78db3edce87c6)
Having a defaulted argument to QUniqueHandle::reset() gives a convenient
way of releasing ownership without assigning a new handle, and makes
the QUniqueHandle API more similar to std::unique_ptr.
Task-number: QTBUG-132507
Change-Id: I842d4545d7cc2da8fe2df08280d0d816ed4be7fd
Reviewed-by: Tim Blechmann <tim.blechmann@qt.io>
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
(cherry picked from commit 248ac4128fe86150532ff7146d0459abb5260946)
Reviewed-by: Qt Cherry-pick Bot <cherrypick_bot@qt-project.org>
(cherry picked from commit 5653829bf8503c38f705893ebbf6b11dd69b211b)
Both whenAll() and whenAny() have two overloads:
* an overload taking two input iterators, e.g.
whenAll(IntpuIt begin, InputIt end)
* an overload taking an arbitrary number of future objects, e.g.
whenAll(Futures &&... futures)
The public APIs are properly constrained, but internally they call
QtPrivate::when*Impl() template functions, that have the same two
overloads, but do not have any constraints.
As a result, passing exactly two QFuture<T>{} objects was leading to
the compiler picking the Impl overload that takes a pair of iterators.
Fix it by applying a subset of constraints from the public API to
the private implementation as well.
Amends 102f7d31c4 which was introduced
for Qt 6.3, so picking down to Qt 6.5.
Fixes: QTBUG-131959
Pick-to: 6.5
Change-Id: Ide29ac9a494d07870e92957c40c1f614e56825f8
Reviewed-by: Mårten Nordheim <marten.nordheim@qt.io>
(cherry picked from commit 8aef5b0d8fd57684abe39c88af8c14d8882ef07b)
Reviewed-by: Qt Cherry-pick Bot <cherrypick_bot@qt-project.org>
(cherry picked from commit 90fab1e4e98962ecaff5fd66d218d4f76cb7b8cd)
We can use hashLengthInternal() to check if the buffer is big enough.
This matches what the QCH::hash() method does, it also has an assert
that `result.size() == ba.size()`, so we can assume this works with
OpenSSL's EVP_MD_get_size() in EVP::finalizeUnchecked().
Amends c70c81b371.
Change-Id: I64935f3d590ab243b361a0b764f011c388820e32
Reviewed-by: Marc Mutz <marc.mutz@qt.io>
(cherry picked from commit b83e825fab16f83f86149ead78efb6ec3d2fa16d)
Reviewed-by: Qt Cherry-pick Bot <cherrypick_bot@qt-project.org>
(cherry picked from commit cc369ad07f9d66fe7ff71170a1a3ddfd2245e8cb)
QSaveFile records past write errors in writeData(), but often the
QFileDevice::writeData() calls it places will succeed because the data
is only being buffered. Instead, the failures are noticed only by
flush(), whose actions do not affect QSaveFilePrivate::writeError.
[ChangeLog][QtCore][QSaveFile] Fixed a bug that caused commit() to
return true and overwrite its intended target file even though it failed
to flush buffered data to the storage, which could cause data loss. This
issue can be worked around by calling flush() first and only calling
commit() if that returns success.
[ChangeLog][QtCore][QSaveFile] Fixed a bug that caused commit() to
return true even after a cancelWriting() call had been placed, if
writing directly to the target file (that is, only with
setDirectWriteFallback() set to true). Note that the state of the file
does not change under those conditions, only the value returned by the
function.
Drive-by clarify a comment from 6bf1674f1e51fd8b08783035cda7493ecd63b44
(Qt 4.6 "Don't drop errors from flush on QFile::close") which had me
chasing the wrong lead.
Fixes: QTBUG-132332
Pick-to: 6.5 5.15
Change-Id: I427df6fd02132d02be91fffd175579c35b9c06cc
Reviewed-by: David Faure <david.faure@kdab.com>
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@gmx.de>
(cherry picked from commit 92373d353cf090faa03cbc8aca505d1784b10b54)
Reviewed-by: Qt Cherry-pick Bot <cherrypick_bot@qt-project.org>
(cherry picked from commit 09d44fdef3acf4146a26d6ae757eb2915b29f147)
It does.
While std::span, does, too, and so users should be using
std::as_const(), it's quite simple to avoid for QSpan, so we'll fix it
in QSpan. This patch adds the reproducer.
Task-number: QTBUG-132133
Change-Id: I2e416fb7344830cd5e0d945cce61491cd6f4a7a5
Reviewed-by: Giuseppe D'Angelo <giuseppe.dangelo@kdab.com>
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
(cherry picked from commit 05b9a4b2deefd586356e1f36d84372b06e74cfe3)
Reviewed-by: Qt Cherry-pick Bot <cherrypick_bot@qt-project.org>
(cherry picked from commit aff082764ae2dc6ca3c7e9c9ee6573bdd5b4a2de)
If there's no way to resolve an actual date with the data parsed, then
the date-text given is invalid, so don't try to fix it up.
Pick-to: 6.5
Fixes: QTBUG-132115
Change-Id: Ic6821bd01394d4dba1be1d25806c372800f8176b
Reviewed-by: Edward Welbourne <edward.welbourne@qt.io>
(cherry picked from commit 72519aeb237a4085aeb6290b0b4088c690fad106)
Reviewed-by: Qt Cherry-pick Bot <cherrypick_bot@qt-project.org>
(cherry picked from commit e2a0ef30da08ed46ea90e2c7aa297c8f8faf0dbb)
We were mostly ignoring them because it looks like most people's options
were longer than their positional arguments. The rest must have just
accepted the enforced wrapping.
But if you have very short options like single-letter only ones or none
at all, the positional argument wrapping is unnecessarily short.
[ChangeLog][QtCore][QCommandLineParser] Made it so the positional
argument descriptions are taken into account in the aligning of text for
helpText().
Fixes: QTBUG-131716
Change-Id: Ib1eee62c7cf4462f6a26fffdec233ba849ebf158
Reviewed-by: David Faure <david.faure@kdab.com>
(cherry picked from commit 8928b0fbb9ca4caf9b63a32b3d2a73a6da096755)
Reviewed-by: Qt Cherry-pick Bot <cherrypick_bot@qt-project.org>
(cherry picked from commit 55a46ec005894394031efbee1b9c2269bb1d8eea)
Like QStringView's, these classes' documentation promised
is_constructible<View,Char[]>, but failed to deliver, because neither
the (const Pointer&) nor the (const Container &) compile for arrays of
unkonwn bounds, and the (const Char*) one is just a QDoc fake.
Apply the same fix as for QStringView: Add a ctor specifically for
arrays of unknown bound, delegating to the (ptr) overload.
The GHS compiler doesn't like the CanConvert static_asserts, so
comment them out for it. The functionality itself is tested by
the from*ArrayWithUnknownSize tests.
[ChangeLog][QtCore][QUtf8StringView/QAnyStringView] Made construction
from arrays of unknown size compile. Such arrays will use the const
Char* constructor, determining the size of the array at runtime.
Pick-to: 6.5
Fixes: QTBUG-112746
Change-Id: I7acdcae3c5bdf80a0bed673e621d53ef34a92a1e
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
(cherry picked from commit 734bd05d0a6d37d6488cf8d1b2b9f79b9329d966)
(cherry picked from commit 51bfc9da41faa41a3161d30879b394a0dac94dcf)
Reviewed-by: Ivan Solovev <ivan.solovev@qt.io>
This is supposed to work, so check it.
Conflict resolutions for 6.8:
- added missing qspan.h include
Change-Id: I201033656f123b09644e5de447cd5d7b038e5155
Reviewed-by: Ivan Solovev <ivan.solovev@qt.io>
(cherry picked from commit 8a3ffe7044249bcfb5185bd87a9713685d48de7b)
Reviewed-by: Qt Cherry-pick Bot <cherrypick_bot@qt-project.org>
(cherry picked from commit 9bcbeea2adeda6b23b030a18af04fce9bf7e5998)
Reviewed-by: Ahmad Samir <a.samirh78@gmail.com>
Adding it here since this test case already has all the infrastructure
for the overload test.
Pick-to: 6.5
Change-Id: I2d7fff9d2d82fed3db2446690a354f939c9a37fc
Reviewed-by: Ivan Solovev <ivan.solovev@qt.io>
(cherry picked from commit 140dac92f2a8ee1f54843f69be4026900a049ee7)
Reviewed-by: Qt Cherry-pick Bot <cherrypick_bot@qt-project.org>
(cherry picked from commit 29e269567610aff4e26738aff507e58e23900133)
Other view tests test Char[N] and const Char[N] separately. Do the
same here.
For many Char types, the array CanConvert test was missing
completely. Add them.
Pick-to: 6.5
Change-Id: I1d0b6de394d548554a547f190e74cb8cead6ecd4
Reviewed-by: Ivan Solovev <ivan.solovev@qt.io>
(cherry picked from commit 0bbbbfc78beb2393ceea0de0fa85942f71d0a2ed)
Reviewed-by: Qt Cherry-pick Bot <cherrypick_bot@qt-project.org>
(cherry picked from commit 443bdb96206a66159ccd07e0704015cf07321327)
The QtDeclarative code causing QTBUG-131906 hits UB, because it tries
to store -666 in an enum {A, B}, which has a valid range of [0,1],
therefore its underlying_type is uint, yet, as per [conv.prom]/3¹,
integer-promotes to _int_ instead, so in Qt 6.8 would cause the
arg(int) overload to be called, outputting -666, while in Qt 6.9, it's
treated (correctly) as an unsigned value, outputting -666's two's
complement, a positive value.
Add a version of the scenario that does not cause UB.
¹ Thanks to Ahmad Samir for digging up the pertinent legalese.
Task-number: QTBUG-131906
Pick-to: 6.5
Change-Id: Iba1a04de523a0b4cd1c87deea40c643cf16df14f
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
Reviewed-by: Fabian Kosmale <fabian.kosmale@qt.io>
(cherry picked from commit e64fd05fecae291c9d7358d2e47d7170995af256)
Reviewed-by: Qt Cherry-pick Bot <cherrypick_bot@qt-project.org>
(cherry picked from commit 06a209384aff389af1b73cc77266f9b392219845)
They are not.
Task-number: QTBUG-112746
Change-Id: I2a20d68543f5914690acbcb4f214b1d98681de6a
Reviewed-by: Ivan Solovev <ivan.solovev@qt.io>
Reviewed-by: Ahmad Samir <a.samirh78@gmail.com>
(cherry picked from commit 67990c16a6b631c1d809d1ad656fc941293b3f83)
Reviewed-by: Qt Cherry-pick Bot <cherrypick_bot@qt-project.org>
(cherry picked from commit 31e30122963796d454d0351da92b0a608f305e50)
The original idea (8b5aa7b6c4) of the
separation of array and pointer ctors was to determine string literal
sizes at compile-time in C++11's limited constexpr semantics.
But when the scanning for NUL characters was added to the array ctor
in 107ff4c1d6, the distinction between
the two ctors became meaningless, because we were able to assume
post-C++11 constexpr to make the Char(0) scan a compile-time action.
Finally, 9e1dc1e8a9 removed the array
ctor in favor of the generic Container ctor. I didn't check whether
the old code handled arrays of unknown size, as in
extern const char16_t str[];
QStringView sv = str;
~~~
const char16_t str[] = "str";
but std::size() (and therefore if_compatible_container) surely bails
on such arrays, and if_compatible_pointer also SFINAEs out. As a
consequence, such arrays cannot be used to construct QStringViews atm.
Fix by adding a new constructor for arrays of unknown size, delegating
to the existing pointer overload.
The GHS compiler doesn't like the CanConvert static_asserts, so
comment them out for it. The functionality itself is tested by
the from*ArrayWithUnknownSize tests.
[ChangeLog][QtCore][QStringView] Made construction from arrays of
unknown size compile. Such arrays will use the const Char*
constructor, determining the size of the array at runtime.
Pick-to: 6.5
Task-number: QTBUG-112746
Change-Id: Ifdb217350d93d38f081c99f14661975491d32076
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
(cherry picked from commit 56faffd92bf0ac459a921ec043a6f3b3dba51acc)
Reviewed-by: Qt Cherry-pick Bot <cherrypick_bot@qt-project.org>
(cherry picked from commit 2dfc0cb6f6f1e29a0a1318793b29b89390543be6)
As documented in the LMDL spec (see comment on unAliasedLinks in the
diff) not all links in the IANA DB are represented as CLDR
aliases. This didn't surface until a recent update to the IANA DB used
symlinks more aggressively.
Add a set of known cases of this to exclude from hasAlternativeName()
checks and let's hope it doesn't grow too often.
Pick to 6.8 required #include <set> that was transitively available on
dev but not on 6.8
Change-Id: Ia12d9fcad96485469ded6bdb7345a400bd34efae
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
(cherry picked from commit ae6b6bf363d71184269104e4bfbb19a9abd27a24)
QChar[N] and ushort[N] are being checked, it's unclear why char16_t[N]
and wchar_t[N] were missing.
Add them.
Pick-to: 6.5 5.15
Change-Id: I9a2df2a75886b950e8c2bdec843e3e693e536f86
Reviewed-by: Ivan Solovev <ivan.solovev@qt.io>
(cherry picked from commit 89401858696d63b8a13c945d5db63856a3b6f5ba)
Reviewed-by: Qt Cherry-pick Bot <cherrypick_bot@qt-project.org>
(cherry picked from commit 67adc2227dc179828d01475c2b9e534bca3152b7)
It can (unlike the other views).
Pick-to: 6.5 5.15
Task-number: QTBUG-112746
Change-Id: Id976429611c53f1c707de1d989c454507b8f4773
Reviewed-by: Ahmad Samir <a.samirh78@gmail.com>
Reviewed-by: Ivan Solovev <ivan.solovev@qt.io>
(cherry picked from commit 01f0305dc6f751d3eb4d1681a2f8f9f3165b547c)
Reviewed-by: Qt Cherry-pick Bot <cherrypick_bot@qt-project.org>
(cherry picked from commit 0c054f6844583136f19608526b0319735d66f0b8)
The lack of this was hidden by other rules (redundant with it) until
CLDR v45, but v46 prunes the redundant rules, breaking this. So
include the missing rule and tweak the code that assumed likely
sub-tag rules preserved language, since this one doesn't. Rework the
tail of withLikelySubtagsAdded() to correctly use this rule, now that
we have it. (The prior comment about there being no match-all was
wrong: CLDR did have it, but our data skipped it.) Amended one test
affected by it (when system locale wasn't en_US).
On picking to 6.8, uiLanguages() needed some coaxing to avoid
duplicate C locale entries in tests of qualified C locale.
Task-number: QTBUG-130877
Change-Id: I2a415b67af4bc8aa6a766bcc1e349ee5bda9f174
Reviewed-by: Mate Barany <mate.barany@qt.io>
(cherry picked from commit 303863170c3ea7d1ee1b7188f507ad432ed3f860)
Conflict resolution for 6.8:
- remove the QEXCEPT_FAIL - 6.8 was never affected
Task-number: QTBUG-131906
Pick-to: 6.5
Change-Id: Icdd647bf6a36ad11e6e19786121d25392b53236c
Reviewed-by: Ivan Solovev <ivan.solovev@qt.io>
(cherry picked from commit 64daf773afff7dbe440621a57f015d985e41f84a)
As previously noted, this backend uses the legacy zone.tab file, which
only gives one territory for each entry, where zone1970.tab gives
several. Until we switch to using the latter, we thus don't know all
the territories in which an IANA ID is used. However, we do have such
data from CLDR and deploy it in the default backend to filter the list
of all available IDs. As a result, the TZ backend lacks some results
of filtering by territory that the base-class would get for it,
because the CLDR data associates more territories with some IDs.
It implements the overload because its lookups in a hash are
potentially more efficient than generating its full list in order to
iterate it while intersecting with a CLDR-derived list of candidates.
Leverage that hash also to do efficient intersecting with the
CLDR-derived list, which lets us avoid duplicates almost for free, to
make the subsequent union with the directly-derived entries from
zone.tab efficient. This incidentally resolves a TODO about handling
of QLocale::AnyTerritory. In the process, add more notes to the
zone{1970,}.tab details.
Restructure the base-class implementations of filtered available IDs
to separate their generation of a CLDR-derived list of candidates from
the intersection of that with the available IDs. For the territory
overload, this lets the TZ backend share the CLDR-derived list; for
symmetry, do the same to the offset-filter, too. (This also makes each
function responsible for fewer things.) Similar treatment may also be
needed for the ICU and chrono::tzdb backends, since the former also
overloads the territory filter and both overload the offset filter.
However, leave those for now (with comments added on the possibility)
until we have evidence they actually do need this.
On picking to 6.8, the chrono::tzdb backend part is gone, as that's
new in 6.9. QTZP_p.h also had a trivially-resolved conflict due to
adjacent 6.9-new content sharing the new protected: block. The initial
pick's conflict in tst_qtimezone.cpp is resolved by picking onto the
correct parent.
Task-number: QTBUG-130877
Task-number: QTBUG-64941
Change-Id: Ibb347df88dd14b55f8f580bb4c9e37e5c56a533a
Reviewed-by: Mate Barany <mate.barany@qt.io>
(cherry picked from commit 54c2d6ded779c3b9843fa18535b79ababa0d1a74)
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
The test was reporting wrong app size values and then
expecting a wrong height after subtracting the system
bar heights. This was happening because an older API
was still being used, using newer APIs fixes that.
Fixes: QTBUG-131338
Pick-to: 6.5
Change-Id: I7306ce62d9c683f84069cc19086a6cd28abf5441
Reviewed-by: Tor Arne Vestbø <tor.arne.vestbo@qt.io>
(cherry picked from commit 50603fddc535179ee02379a8807fe5cb104e0aae)
Pick-to: 6.5
Change-Id: Iea06638d1c8eebf36ae010dad23fbeab4b890597
Reviewed-by: Petri Virkkunen <petri.virkkunen@qt.io>
(cherry picked from commit ebdc01c1861667116aa09914b46df1ac8504baf9)
Reviewed-by: Tor Arne Vestbø <tor.arne.vestbo@qt.io>
The RecursiveList structure was defined like this:
struct RecursiveList : public QList<RecursiveList> {};
However, this definition does not make any sense when it comes to the
relational operators. QList<T> needs to have operator<() defined for
the contained type T.
The current implementation of QTypeTraits::compare_lt_result_container
simply enables operator<() if Container is a base type of the contained
type. This unblocks the compilation of checks like
static_assert(QTypeTraits::has_operator_less_than_v<RecursiveList>);
However, this is useless in practice.
This commit updates the struct to have a meaningful definition of
operator<().
Amends 9f13842fe6.
Task-number: QTBUG-120305
Pick-to: 6.5
Change-Id: Iaee96385a33ff131bdceabe945265b3285a370c2
Reviewed-by: Marc Mutz <marc.mutz@qt.io>
(cherry picked from commit 58c492a4dc9868faff7e9334758947cf0027eeac)
Reviewed-by: Qt Cherry-pick Bot <cherrypick_bot@qt-project.org>
There was a copy-paste error in NoCmpRecursiveList definition.
Amends 9f13842fe6.
Pick-to: 6.5
Change-Id: I1487710e51a8b92c53b79b747c324654d9b9f334
Reviewed-by: Marc Mutz <marc.mutz@qt.io>
(cherry picked from commit 65fede8fae459da86784b0757ca6c3179a8fb883)
Reviewed-by: Qt Cherry-pick Bot <cherrypick_bot@qt-project.org>
The AnyTerritory entries in the zoneDataTable are derived from
territory="ZZ" entries in the upstream CLDR data; the World ones from
territory="001". The latter give the default IANA ID for each MS ID,
the former give an (often legacy) IANA ID for the MS ID, that is not
based on geography. Some of these are being removed at CLDR v46.
The documentation said the ZZ entries have "no known territorial
association", hinting that there may be some (unknown) territorial
association; however, CLDR's inclusion of them is as entries with a
known non-territorial association, so revise the phrasing to reflect
this.
Also document that windowsIdToDefaultIanaId() returns empty when
there is no territory-specific value, and callers can use the
territory-neutral call to get a suitable value in that case. (They
may, however, wish to distinguish this case, to treat it differently,
so I decided not to just return that in place of empty in any case.)
The upstream CLDR tables do have entries for territory 001, so we
should report these if asked for World as territory. Amend the
available zone ID lookup and mapping from MS to IANA functions that
take a territory to duly handle World via the default-data that was
derived from 001 data in CLDR, instead of from the territory-varying
table, from which those were effectively filtered out when generating
the two tables. Update docs to mention this handling of World, for
contrast with that of AnyTerritory.
In the process remove a spurious split-on-space from the MS to default
IANA lookup, asserting there is no space (in a field now stored in the
table for single IANA ID entries, instead of the one for space-joined
lists of them in which it used to be stored, before I noticed it's
always only one ID). There is a matching assertion in the cldr.py code
that extracts the data. Added an assertion to this last, that each
default IANA ID given by CLDR's MS data does in fact also appear as
one of the IANA IDs for at least one territory (potentially ZZ), and
comment in C++ code on why this means we don't need to scan the
windowsDataTable in a few places, where it would just produce
duplicate entries.
On picking to 6.8, removed the timezone_locale addition, only relevant
on 6.9 and later.
[ChangeLog][QtCore][QTimeZone] Corrected handling of QLocale::World
and clarified in docs how QLocale::AnyTerritory is handled when
QTimeZone selects zones by territory.
Task-number: QTBUG-130877
Change-Id: I861c777c68b0cb73a194138fe23fbff839df49e6
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
(cherry picked from commit e23dc7c420297fb62db9834a17c59bbf5992dad7)
Reviewed-by: Mate Barany <mate.barany@qt.io>
We want to avoid caling toMSecsSinceEpoch() since it's expensive for
LocalTime (which is presumed to be the common case). We can do so when
both sides have the same offset from UTC (and this can cheaply be
determined) but that's no help for two local times months apart, one
in DST the other not. However, in this case, the difference in millis
is big enough that no plausible difference in offset can overcome it,
so we can again avoid toMSecsSinceEpoch() and simply compare millis.
This should make some previously-expensive comparisons cheap.
Add test-cases to the QDateTime ordering test that verify this doesn't
lead to mis-comparison at the biggest offset-difference known.
Fixes: QTBUG-131491
Change-Id: I1afd5d058c8663c908f898d4c50d0837549b87db
Reviewed-by: Christian Ehrlicher <ch.ehrlicher@gmx.de>
(cherry picked from commit ef540d77751e24fe0b345694f43cdafca3434c68)
Reviewed-by: Qt Cherry-pick Bot <cherrypick_bot@qt-project.org>
87d8ee755b changed the logging behavior
on Android to use the category as tag. But it was missed that
`formatLogMessage` only fills the fields given in the messagePattern.
Our default messagePattern includes the category but on Android it
is never filled since this change.
So we change the default messagePattern on Android to omit the category
and only use the category as tag when the category is not given in the
messagePattern.
Pick-to: 6.5
Task-number: QTBUG-94708
Change-Id: I80f65d0f7f8c0ca9c2fff2dcd63d4599848b6e2b
Reviewed-by: Assam Boudjelthia <assam.boudjelthia@qt.io>
(cherry picked from commit 709bc29a1d2abf40bf3f1c067fe3b96b6e0c09d0)
Reviewed-by: Qt Cherry-pick Bot <cherrypick_bot@qt-project.org>
A recent update [0] to the IANA timezone database has improved the
historical accuracy of transitions in Mexico. It turns out that the
transition at the start of 1970 was not a real event. The only actual
change then was that Mazatlan has followed the main Mexico zone rules
since then, having ignored some transitions of the main zone in
earlier decades.
[0] 812aff32b3
Picking to 6.8 got a trivially-resolved conflict due to dev having one
more BackendKludges flag to take into account. Removed 6.5 from the
destinations for picking because the removed test was added since
then.
Pick-to: 6.8.1
Fixes: QTQAINFRA-6757
Change-Id: I65d8ea1dfa76ce635bf11ffcc1ded8c9ceaf9fec
Reviewed-by: Paul Wicking <paul.wicking@qt.io>
(cherry picked from commit 8a27c310b1e4ebc74999e09551ca5eab4570a68e)
Commit 1ed0dd88a3 ("QThread/Unix: make
QThreadPrivate::finish() be called much later") introduced this problem.
Commit 4fabde349f16b59f37568da2a4c050c6dd53a34e split the thread
termination in two phases, but did not fix this.
This re-applies commit 950b35cf97 ("Clear
the current thread data for the main thread"), which was reverted in
commit 7dc622290b ("Make sure QThreadData
and QAdoptedThread object is destroyed at app exit"), both from Qt 5.1.
Between Qt 5.1 and 6.7, the responsibility of clearing the
QAdoptedThread for the main thread was split: it could occur either in
~QCoreApplicationData if exit() was called in that thread or in
~QThreadData() if it wasn't (e.g., when the Qt "main thread" is not
main()'s thread):
* frame #0: 0x0000000101db8a28 QtCore`QAdoptedThread::~QAdoptedThread(this=0x000060000176c070) at qthread.cpp:139:1
frame #1: 0x0000000101db81eb QtCore`QThreadData::~QThreadData(this=0x0000600002468000) at qthread.cpp:82:5
frame #2: 0x0000000101db8379 QtCore`QThreadData::~QThreadData(this=0x0000600002468000) at qthread.cpp:57:1
frame #3: 0x0000000101db841c QtCore`QThreadData::deref(this=0x0000600002468000) at qthread.cpp:108:9
frame #4: 0x0000000101f4ec79 QtCore`destroy_current_thread_data(p=0x0000600002468000) at qthread_unix.cpp:104:11
This commit centralizes and gives ~QThreadData() the exclusive
responsibility. That requires not resetting QThreadData::threadId so
~QThreadData can know it is theMainThread.
Fixes: QTBUG-130895
Task-number: QTBUG-129927
Task-number: QTBUG-129846
Task-number: QTBUG-130341
Task-number: QTBUG-117996
Change-Id: Ie3f3cbdc5523837b505cfffd95fba5e6498b5069
Reviewed-by: Ivan Solovev <ivan.solovev@qt.io>
(cherry picked from commit 65093a84c2b94b1543fd4593bc45d491951d28d4)
When we consolidated "all" test functions to use RUN_TEST_FUNC, the
cancel() test function was inexplicably overlooked.
This patch ports this last test function.
Amends 4c00337ccb8e4266fa5a4af4fba40e5b62aba81b, whose commit message
contains rationale for this change, but not repeated here.
Pick-to: 6.5 6.2
Change-Id: I9461f8e816a25c731d6229df15ee88dcc8c4e165
Reviewed-by: Ivan Solovev <ivan.solovev@qt.io>
(cherry picked from commit 3e71f256d5813ef699a4030f718fee1d1e53bba1)
Reviewed-by: Qt Cherry-pick Bot <cherrypick_bot@qt-project.org>
QWinRegistryKey::value<T> returns an optional<T> which can make code a
bit more readable than the std::pair<DWORD, bool> return value from
QWinRegistryKey::dwordValue.
This implements part of the "TODO: Remove once all usages are migrated
to new interface." comment in QWinRegistry class because it allows us to
remove the QWinRegistryKey::dwordValue function.
Change-Id: If568de4e31778e91ce7aadadb4aac90e41222826
Reviewed-by: Edward Welbourne <edward.welbourne@qt.io>
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
(cherry picked from commit 7a63a25ef71fbbe7b3eaa6ecb9a26c3e497e582a)
Reviewed-by: Qt Cherry-pick Bot <cherrypick_bot@qt-project.org>
The implementation checks the state and returns without storing the
exception when its canceled or finished.
Add tests for both situations.
Done-with: Ivan Solovev <ivan.solovev@qt.io> (analysis and phrasing)
Pick-to: 6.5 6.2
Fixes: QTBUG-128405
Change-Id: I4610a022ea12e1bc9ce24cb17b972b5b9e051f0a
Reviewed-by: Ivan Solovev <ivan.solovev@qt.io>
(cherry picked from commit 6efec3850da188d3bba075185aa6e5c264c815eb)
Reviewed-by: Qt Cherry-pick Bot <cherrypick_bot@qt-project.org>
Don't just silently succeed.
As a drive-by, keep the skipped code visible to the compiler.
QThread::create() is universally available since
4a5f3c8b93, so the skipped code ought to
compile.
Amends ac6a9f9bfa.
Change-Id: I193ecf802ffbfd103747aa34eb307fb1a814a94e
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
(cherry picked from commit 4e1b20893aeeb9580bc6b52ac4baea297675bfee)
Reviewed-by: Qt Cherry-pick Bot <cherrypick_bot@qt-project.org>
Commit 1ed0dd88a3 moved the finish()
functionality from immediately after run() returns to the time of
thread-local destruction, to make sure that user destructors didn't run
after our cleaning up. But as a side effect, it made other user code run
too late, after some thread-local statics had been destroyed.
This is a common practice, which causes the destructor for worker to run
too late:
worker->moveToThread(thread);
...
QObject::connect(thread, &QThread::finished, thread, &QObject::deleteLater);
QObject::connect(thread, &QThread::finished, worker, &QObject::deleteLater);
This commit splits the cleanup in two phases: QThreadPrivate::finish(),
which runs immediately after run() and will call back out to user code
(finished() signal and delivery of deleteLater()), and cleanup() that
cleans up the QThread{Private,Data} state and destroys the event
dispatcher. That destruction is the only call out to user code.
I've removed the complex mix of pre-C++11 pthread_setspecific() content
and C++11 thread_local variables in favor of using one or the other, not
both. We prefer the thread-local for future-proofing and simplicity, on
platforms where we can verify this C++11 feature works, and because it
allows us to clean up QThreadData and the event dispatcher as late as
possible. (There's some code that runs even later, such as pthread TLS
destructors, used by Glib's GMainLoop)
Unfortunately, we can't use it everywhere. The commit above had already
noticed QNX has a problem and recent bug reports have shown other
platforms (Solaris, MUSL libc) that, 13 years after the ratification of
the standard, still have broken support, so we use pthread for them and
we call cleanup() from within finish() (that is, no late cleaning-up,
retaining the status quo from Qt 4 and 5). See QTBUG-129846 for an
analysis.
Drive-by moving the resetting of thread priority to after finished() is
emitted.
[ChangeLog][QtCore][QThread] Restored the Qt 6.7 timing of when the
finished() signal is emitted relative to the destruction of thread_local
variables. Qt 6.8.0 contained a change that moved this signal to a later
time on most Unix systems, which has caused problems with the order in
which those variables were accessed. The destruction of the event
dispatcher is kept at this late stage, wherever possible.
Fixes: QTBUG-129927
Fixes: QTBUG-129846
Fixes: QTBUG-130341
Task-number: QTBUG-117996
Change-Id: Ie5e40dd18faa05d8f777fffdf7dc30fc4fe0c7e9
Reviewed-by: Edward Welbourne <edward.welbourne@qt.io>
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
(cherry picked from commit 4fabde349f16b59f37568da2a4c050c6dd53a34e)
The commit 1f22fc995a introduced nested
functions without properly checking their return status with
QTest::currentTestFailed().
Later, commit 855c448469 used the same
pattern.
Fix all the affected tests by wrapping the function calls into the
pre-existing RUN_TEST_FUNC macro.
Both commits were picked to 6.5, so do the same with this patch.
Pick-to: 6.5
Change-Id: I3e11bf5724d9b89c75a4e5c73c96f9566c36b5cd
Reviewed-by: Marc Mutz <marc.mutz@qt.io>
Reviewed-by: Edward Welbourne <edward.welbourne@qt.io>
(cherry picked from commit 4c00337ccb8e4266fa5a4af4fba40e5b62aba81b)
Reviewed-by: Qt Cherry-pick Bot <cherrypick_bot@qt-project.org>
For some reason, 4d6cf54664a98e01d4caab42431c62d6deb9f86e didn't treat
all lambdas, and no-one noticed.
This patch completes the task by transforming the remaining lambdas to
be non-move-only.
Amends 4d6cf54664a98e01d4caab42431c62d6deb9f86e, whose commit message
is pertinent to this patch, too, but not repeated here.
Pick-to: 6.5 6.2
Change-Id: I6dfe6eb438065e16fcb7ab1f2efcb64c0146bb6c
Reviewed-by: Ivan Solovev <ivan.solovev@qt.io>
(cherry picked from commit 5c3dc2ecb97b69eb12e06c499ddb50666be60be6)
Reviewed-by: Qt Cherry-pick Bot <cherrypick_bot@qt-project.org>
Those files are read by reuse to complement or override the copyright
and licensing information found in file.
The use of REUSE.toml files was introduced in REUSE version 3.1.0a1.
This reuse version is compatible with reuse specification
version 3.2 [1].
With this commit's files,
* The SPDX document generated by reuse spdx conforms to SPDX 2.3,
* The reuse lint command reports that the Qt project is reuse compliant.
[1]: https://reuse.software/spec-3.2/
Task-number: QTBUG-124453
Task-number: QTBUG-125211
Change-Id: I01023e862607777a5e710669ccd28bbf56091097
Reviewed-by: Edward Welbourne <edward.welbourne@qt.io>
Reviewed-by: Joerg Bornemann <joerg.bornemann@qt.io>
(cherry picked from commit 4c83657448f66ac0ee46900aff2e5a230eda49b0)
VxWorks, at least as used in the 6.8 CI, has a problem with move-only
lambdas passed to QThread::create(). This might have been the reason
for the resurrection of the cxx11_future feature in
408430a543ef605157963b1a894847c7ba9f9956.
This patch tries to work around the issue by not moving the promise
into the lambda capture, but caputuring it by reference, and then
moving it into a local stack variable. In all cases, the ThreadWrapper
is created after the promise, so, since the ThreadWrapper dtor join()s
(like jthread), the promise out-lives the thread, and the destruction
of the promise happens-after the join() with the thread. So, no data
races. Code mustn't touch the promise object from the main thread once
the ThreadWrapper has started, but that's the case already and the
test functions are small enough for any such future use to be obvious
in review.
Amends 385f0732d9 and
4e1b20893aeeb9580bc6b52ac4baea297675bfee.
It appears that dev no longer has the problem¹, but apply this to dev,
too, to maintain consistency with the test in last three LTS releases.
¹ 4e1b20893aeeb9580bc6b52ac4baea297675bfee passed, but its cherry-pick
to 6.8 did not
Pick-to: 6.5 6.2
Change-Id: If1feba9d3a7f8677789d45fa9f4990cf44145709
Reviewed-by: Ivan Solovev <ivan.solovev@qt.io>
(cherry picked from commit 4d6cf54664a98e01d4caab42431c62d6deb9f86e)
Reviewed-by: Qt Cherry-pick Bot <cherrypick_bot@qt-project.org>
They are an unnecessary hit on build time.
Rewrite the tests using the standard Qt conventions.
Change-Id: Iac47e62c58b1805c3b4a0ac4f4b3db206c66cc65
Reviewed-by: Axel Spoerl <axel.spoerl@qt.io>
(cherry picked from commit 6ddf74716d7b2fe0df5dd9c0172f9ecf936d5c5d)
Reviewed-by: Qt Cherry-pick Bot <cherrypick_bot@qt-project.org>
There's nothing in our documentation saying this is permitted, but we
have specific code for it on Windows and it works on Unix because we
just wait on a QWaitCondition.
Change-Id: Id6331fa9aad473cb4f35fffdf8bb04d9a34cd108
Reviewed-by: Mårten Nordheim <marten.nordheim@qt.io>
Reviewed-by: Ivan Solovev <ivan.solovev@qt.io>
Reviewed-by: Edward Welbourne <edward.welbourne@qt.io>
(cherry picked from commit 5b5297fe87859f59a7aaf5e86a8915c00714fefa)
Reviewed-by: Qt Cherry-pick Bot <cherrypick_bot@qt-project.org>
Lambdas that only capture by reference never need to be mutable.
Drop the keyword (and the now-superfluous parentheses).
Amends 385f0732d9.
Pick-to: 6.5 6.2
Change-Id: Iaa6451c1c26f1aed198c447ee59198c4e0285b15
Reviewed-by: Mårten Nordheim <marten.nordheim@qt.io>
(cherry picked from commit 37daf732d5089c58a6428124263243509d152987)
Reviewed-by: Qt Cherry-pick Bot <cherrypick_bot@qt-project.org>
Volker Hilsheimer
Volker Krause
Marc Mutz
Jøger Hansegård
Ivan Solovev
Ahmad Samir
Thiago Macieira
Edward Welbourne
Assam Boudjelthia
Lars Schmertmann
Lucie Gérard
Friedemann Kleint