| Commit message (Collapse) | Author | Age | Files | Lines |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
The old code is just plain wrong for negative julian days. Replaced
with plain math from The Calendar FAQ [1], which is correct for all
julian days, provided you use mathematical integer division (round to
negative infinity) rather than c++11 integer division (round to zero).
[1] http://www.tondering.dk/claus/cal/julperiod.php
While the conversion code works for up to around JD +/- (2^63/4), we
only use an int for the year in the API, so this patch limits minJd()
and maxJd() to 1 Jan (2^31) BC and 31 Dec (2^31-1) AD, respectively.
Note that while the new conversion code looks like it would be more
expensive than the old, gcc will in fact be able to optimize it to be
slightly faster (probably because x86 hardware implements round to
negative infinity, and so GCC manages to optimize floordiv to a single
instruction, compared to the three instuctions needed for operator/).
In the following test application, run with a release mode Qt and
redirecting stderr to /dev/null, I measured an improvement from
6.81s +/- 0.08s to 6.26s +/- 0.16s user time over five runs on an
otherwise idle x86_64 system.
int main(int, char *[])
{
int year, month, day;
qint64 jd;
for (qint64 i = Q_INT64_C(-1048576) ; i < Q_INT64_C(1048576); ++i) {
QDate::fromJulianDay(i).getDate(&year, &month, &day);
jd = QDate(year, month, day).toJulianDay();
qDebug() << jd << year << month << day;
}
}
Change-Id: Ifd0dd01f0027f260401f7f9b4f1201d2b7a3b087
Reviewed-by: David Faure (KDE) <faure@kde.org>
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
Reviewed-by: Lars Knoll <lars.knoll@digia.com>
|
|
|
|
|
| |
Change-Id: Ic521d9d2ffb1b8e3b14d9cebdeb3dc7a5e08580e
Reviewed-by: Jędrzej Nowacki <jedrzej.nowacki@digia.com>
|
|
|
|
|
|
|
|
| |
Task-number: QTBUG-27391
Change-Id: I68b37ffa645be21d4d23b205bc052540b9aba7f4
Reviewed-by: Olivier Goffart <ogoffart@woboq.com>
Reviewed-by: João Abecasis <joao@abecasis.name>
|
|
|
|
|
|
|
|
|
|
| |
hasUnquotedAP currently only checks for an a or A, which is wrong
according to both the toString documentation and the comments for
hasUnquotedAP.
Change-Id: I03015734b846fe761085cf8f8fca2b29210cff97
Reviewed-by: Jon Severinsson <jon@severinsson.net>
Reviewed-by: Jędrzej Nowacki <jedrzej.nowacki@digia.com>
|
|
|
|
|
|
|
|
|
|
| |
We can insert directly on the most left-most Node.
We always enforce an insert here (unlike the insert call),
but that is not a problem since the keys in a std::map are unique.
Change-Id: Ib409b90ffc57a5a43dab4a4b08d34f6fdabd057f
Reviewed-by: Lars Knoll <lars.knoll@digia.com>
|
|
|
|
|
|
|
|
|
|
| |
Following the Code Style described in
http://qt-project.org/wiki/Qt_Coding_Style#e289ee44592e9c32d4212069f0806daf
There is no need for a 'break' after a 'return'.
Change-Id: I1eca350391a7e4e14e504d60b24b69982cc5ac47
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
This suggestion keeps track of the most left node.
The point is that constBegin() becomes a lot faster.
That speeds up iteration a bit, and makes it O(1) to get the
first element. The penalty in insert and remove is very small.
On large trees it seems to be less than 1%.
It should be noticed that constBegin() is a very common hint
on my planned change to 5.1, and this opperation will without
this patch cost 2 x log N. One when the user calls the hint
with begin - and one where it is compared with begin.
Other std::maps has a very fast begin(). E.g
http://www.cplusplus.com/reference/stl/map/begin/
(begin with constant time)
Change-Id: I221f6755aa8bd16a5189771c5bc8ae56c8ee0fb4
Reviewed-by: Lars Knoll <lars.knoll@digia.com>
|
|
|
|
|
|
|
|
|
| |
Giving the std-map a hint (normally) improves insert performance.
There seems to be no reason not to provide this hint.
Change-Id: I4344607ebf54574a3ae9666d87a41a3c14762361
Reviewed-by: Robin Burchell <robin+qt@viroteck.net>
Reviewed-by: Lars Knoll <lars.knoll@digia.com>
|
|
|
|
|
|
|
|
| |
This one was missed when the QDate range was extended.
Change-Id: I0dbcc9fdebca88f7397203d8e539429dcff9ac30
Reviewed-by: Lars Knoll <lars.knoll@digia.com>
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
|
|
|
|
|
|
|
|
|
|
| |
We had this issue in Qt 4 before: Clang considers the use of d uninitialized
in the initializer list for d. This might or might not be legal, the clang
devs are divided on that topic. Let's just use a more common form and ignore
the issue language lawyers.
Change-Id: I3324255963a6d0f4a9057fe0d1d9ae868d3efef7
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
|
|
|
|
|
|
|
|
|
| |
This is in line with what QTime::addMSecs and QDate::addDays do, for
example.
Change-Id: I902112486727f74f669a97bde6c42028e0838f8d
Reviewed-by: Jon Severinsson <jon@severinsson.net>
Reviewed-by: Jędrzej Nowacki <jedrzej.nowacki@digia.com>
|
|
|
|
|
|
|
| |
Change-Id: If47de3dc047ac4f8a4a1498cf225e03bbbf4110e
Reviewed-by: Mitch Curtis <mitch.curtis@digia.com>
Reviewed-by: Jędrzej Nowacki <jedrzej.nowacki@digia.com>
Reviewed-by: Jon Severinsson <jon@severinsson.net>
|
|
|
|
|
|
|
|
|
|
|
|
| |
Some statements could not be tested, such as default cases of switches
where all possible cases are already handled and some statements where
the system locale is used.
I also removed some statements that would never be reached and hence
will never be able to be covered by tests.
Change-Id: I8ea3071f66d942d986e65708732af6751d36b5e3
Reviewed-by: Jędrzej Nowacki <jedrzej.nowacki@digia.com>
|
|
|
|
|
|
|
|
|
|
| |
Converting the date 1 January 0001 to a string in the format
"dd MMMM yyyy" currently results in the string "01 January 1", but it
should be "01 January 0001".
Task-number: QTBUG-27287
Change-Id: Ia025da7343d1c36aaee34c47c6db551a6e89595f
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
remove() can use non-detaching iterators internally before calling
erase(), which hasn't been exploited so far, so that the detach() in
erase() never actually detached. When using erase() from outside,
you can't do it legally without calling begin() or end() that detach()
before erase() is called.
Now remove() doesn't detach anymore, and detaching in erase() works.
With new tests that fail after changing only the erase() callers
and pass again after fixing erase().
Change-Id: I47c0a9e362dce8628ec566f5437d951755de96c8
Reviewed-by: Thorbjørn Lund Martsum <tmartsum@gmail.com>
Reviewed-by: Olivier Goffart <ogoffart@woboq.com>
Reviewed-by: Konstantin Ritt <ritt.ks@gmail.com>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
These containers don't make sense and will just result in no action
being taken (all items added will simply be discarded), but it
shouldn't crash due to a division by zero.
Update the documentation to explain the fact.
Task-number: QTBUG-27339
Change-Id: Ib9acf5c0a9a826e6853e7beaf5e56511fde98dc6
Reviewed-by: Giuseppe D'Angelo <giuseppe.dangelo@kdab.com>
Reviewed-by: Ian Walters <ian@walters.id.au>
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
The IA-64 C++ ABI does not encode the return type for non-template
functions (QVector is the template, not the function), which means that
these two functions have the same signature:
Node *QVector<Node>::begin()
typename class QTypedArrayData<Node>::iterator QVector<Node>::begin()
[both are _ZN7QVectorI4NodeE5beginEv]
When linking compilation units compiled with different
QT_STRICT_ITERATORS settings, only one of the two out-of-line copies
will survive. Depending on the ABI, we may have a problem: the ABI can
say that a function returning a structure takes an implicit first
parameter, which a function returning a regular pointer doesn't.
Task-number: QTBUG-27277
Change-Id: I57a59e5a7c46f55faabfe85c073dca89d2a7bbf3
Reviewed-by: Jan Kundrát <jkt@flaska.net>
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
|
|
|
|
|
|
|
| |
Change-Id: Ie107e451a7b2a084fb6131939f212acd412a9df6
Reviewed-by: Mitch Curtis <mitch.curtis@digia.com>
Reviewed-by: Jerome Pasion <jerome.pasion@digia.com>
Reviewed-by: Jan Arve Sæther <jan-arve.saether@digia.com>
|
|
|
|
|
|
|
|
| |
They are introduced in PCRE 8.31.
Change-Id: Id0447b381d5e23996d4e87ae0368b07a8bc1c318
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@digia.com>
|
|
|
|
|
| |
Change-Id: Ibacd8f260950db195a17cffe8396e8460f227eb9
Reviewed-by: Jan Arve Sæther <jan-arve.saether@digia.com>
|
|
|
|
|
|
|
|
| |
Documentation has been updated to reflect changes in Qt5.
Change-Id: I3d54d1875962bd27c43bb360ae7b3fda0b7702ba
Reviewed-by: Andreas Holzammer <andreas.holzammer@kdab.com>
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Add BoundaryReason::BreakOpportunity flag that will be returned by the
boundaryReasons() when the boundary finder is at the break opportunity
position that might be not an item boundary.
This is the same as (StartWord || EndWord) in Grapheme and Sentence modes;
in Word and Line modes, BreakOpportunity flag might occur between the words
or in between of Line boundaries (e.g. in conjunction with SoftHyphen flag).
In other words, the text boundaries are always break opportunities, but not vice versa.
StartWord and EndWord flags has been deprecated by new StartOfItem and EndOfItem
flags which are not about the word boundaries only. In line breaking,
StartOfItem and EndOfItem are set for the mandatory breaks only.
Change-Id: I79bf297e2b988f5976f30cff0c8ca616385f6552
Reviewed-by: Konstantin Ritt <ritt.ks@gmail.com>
|
|
|
|
|
|
|
|
|
|
| |
After allocating memory on the heap it is ATM not possible to use the
memory on the stack again, QVarLengthArray then uses/resizes only
the memory on the heap. But the memory on stack could be used
if it is big enough.
Change-Id: I566003c25dd1093eb6ba8087a1e5378a11712934
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
After a exception is thrown in resize(), QVarLengthArray has an
invalid state with ptr == 0. On the next resize call when malloc
returns a valid pointer, oldPtr is 0 and it could crash in
memcpy because the pointer to the source is 0.
The patch ensures the valid pointer isn't overwritten with NULL.
When exceptions are disabled the user must ensure that malloc will
not fail.
Change-Id: Id12a5e3e1eacc551e4d1b64cba8e8414cfebd6e1
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
|
|
|
|
|
|
|
|
|
|
|
| |
Any file including qstringbuilder.h would trigger a warning when
compiled with QT_ASCII_CAST_WARNINGS defined since it implicitely
converts a QString to QByteArray.
Explicitely call toUtf8() to fix the issue.
Change-Id: If20f9d4571c5d1ed789564196c9f1331e1efd1d9
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
|
|
|
|
|
|
|
|
|
|
| |
that will be returned by boundaryReasons() when the boundary finder
is at the line end position (CR, LF, NewLine Function, End of Text, etc.).
The MandatoryBreak flag, if set, means the text should be wrapped at a given position.
Change-Id: I32d4f570935d2e015bfc5f18915396a15f009fde
Reviewed-by: Konstantin Ritt <ritt.ks@gmail.com>
Reviewed-by: Lars Knoll <lars.knoll@digia.com>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
When QSharedPointer is created with no user-specified custom deleter,
instead of storing a pointer in ExternalRefCount::destroyer to a
static function which, in turn, calls normalDeleter<T> indirectly (via
another function pointer), specialise the CustomDeleter class and make
it not store the pointer, but instead do the deleting directly.
The benefits are:
- the QSharedPointer's private data is smaller
- there is no double-indirection via indirect jumps to the actual
deleter
Change-Id: Ice5653c144912efb1226e432267a047b9799aaca
Reviewed-by: Olivier Goffart <ogoffart@woboq.com>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
We don't need two ints if we can do it with one. We need to store only
the clock type to be used, which is either CLOCK_MONOTONIC or
CLOCK_REALTIME. In addition, we need an invalid value to indicate that
the test hasn't been done yet.
This also gets rid of an atomic that wasn't QBasicAtomicInt.
Disassembly of the compiled code indicates this is the most optimal
solution:
- one single variable load
- one test-and-branch to verify that a test is needed
- straight call to clock_gettime afterwards
Change-Id: I49788e45b50b6b7fcb7ba4e9332e5352a4428933
Reviewed-by: Olivier Goffart <ogoffart@woboq.com>
|
|
|
|
|
|
|
|
|
| |
No need to check for overflows since the change to qint64. as less than
half the qint64 range is a valid julian day, any overflow will be
detected as an invalid date anyway.
Change-Id: I3b6cad24e245ed9418c5804484f846b0b692153a
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
|
|
|
|
|
|
|
|
|
|
| |
addMSecs() / msecsTo() have always used qint64, and when QDate was changed
to use a 64-bit julian day, QDateTime::addDays() and QDateTime::daysTo() was
changed to use qint64 in order to support the full extended range, but
addSecs() and secsTo() seems to have been forgotten.
Change-Id: I3acc35ee2bcc9f353650eb42f97d428f706b2db6
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Version 6.2 of the Unicode Standard is a special release
dedicated to the early publication of the newly encoded Turkish lira sign.
In addition, there are some significant changes to the Unicode algorithms
for text segmentation and line breaking to improve breaking for emoji symbols.
For more details, see http://www.unicode.org/versions/Unicode6.2.0/
Change-Id: I21cfd4f307e41b41a19d36cce87f7a44c2661bc2
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
Reviewed-by: Lars Knoll <lars.knoll@digia.com>
|
|
|
|
|
|
|
|
|
| |
QMap has 4 different maximumNode() functions, but they are all
unused and just in the way.
Change-Id: I012145cfae32898830ec95e229514d9be838ae2f
Reviewed-by: Lars Knoll <lars.knoll@digia.com>
Reviewed-by: Robin Burchell <robin+qt@viroteck.net>
|
|
|
|
|
|
|
|
|
|
|
| |
The test is useless as we assert if the requested size exceeds
a certain limit. We could, as an alternative,
throw an exception, but in the end it's the caller's responsibility
to ensure that the requested size is a sane value.
Task-number: QTBUG-27285
Change-Id: I738950a6a2b51671a54e4d25c7e4c3ac0d7f63b8
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
|
|
|
|
|
|
|
| |
Task-number: QTBUG-22967
Change-Id: I604e6a725d46eab4c4369ebb54e8c9ea1350f492
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
|
|
|
|
|
|
|
| |
for the case when the boundary finder is assigned to an invalid one.
Change-Id: I5b60984ff3fd99972fcae21895684bd83b012780
Reviewed-by: Simon Hausmann <simon.hausmann@digia.com>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Functions like QByteArray::at() assert the given index:
Q_ASSERT(i >= 0 && i < size();
These functions typically get inlined. Now if the index is
e.g. size() - 2, then gcc will emit an ugly warning in
client code ("assuming signed overflow does not occur when assuming
that (X - c) > X is always false").
This can be easily prevented by casting both sides of the second
comparison in the assertion to their unsigned type. The explicit
comparison to zero is then no longer necessary, since that condition
is tested implicitly by the other comparison due to unsigned arithmetic.
Change-Id: Ic7244e1fa5da00a47d1fe0ed56fb81c23d444dfe
Reviewed-by: hjk <qthjk@ovi.com>
Reviewed-by: Marc Mutz <marc.mutz@kdab.com>
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
|
|
|
|
|
| |
Change-Id: I9dccb9f64578cf8c9da50524c5d4d1201ab488bd
Reviewed-by: Lars Knoll <lars.knoll@digia.com>
|
|
|
|
|
|
|
|
| |
It's perfectly correct to leave members uninitialised, since they are zero initialized.
Change-Id: I0d0c737cf35793a2633d44ce194af7f489903c03
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
Reviewed-by: Jędrzej Nowacki <jedrzej.nowacki@digia.com>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
A simple heuristic is used to detect the word beginning and ending by
looking at the word break property value of surrounding characters.
This behaves better than the white-spaces based implementation used before
and makes it possible to tailor the default algorithm for complex scripts.
BIG FAT WARNING: The QCharAttributes buffer now has to have a length
of string length + 1 for the flags at end of text.
Task-Id: QTBUG-6498
Change-Id: I5589b191ffde6a50d2af0c14a00430d3852c67b4
Reviewed-by: Konstantin Ritt <ritt.ks@gmail.com>
|
|
|
|
|
|
|
| |
Negation operator ("!") have precedence over bitwise and ("&").
Change-Id: I39e2d99da6eaa4477bbe35a1259f745e05c9841a
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
|
|
|
|
|
| |
Change-Id: Ia73501e1e8a562cfab750d1796fb597cd89eab74
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
|
|
|
|
|
|
|
|
| |
Change copyrights and license headers from Nokia to Digia
Change-Id: If1cc974286d29fd01ec6c19dd4719a67f4c3f00e
Reviewed-by: Lars Knoll <lars.knoll@digia.com>
Reviewed-by: Sergio Ahumada <sergio.ahumada@digia.com>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Introduce QCharAttributes and use it instead of HB_CharAttributes everywhere in Qt
(in Harfbuzz, the HB_CharAttributes is only used in the text segmentation algorithm
which has been moved from HB to Qt (well, most of it)).
Rename some members to better reflect their meaning,
remember to keep HB_CharAttributes in sync with QCharAttributes.
Also replace HB_ScriptItem with a (temporary) QUnicodeTools::ScriptItem struct
that will be replaced with a more efficient/friendly solution a bit later.
The soft hyphen and the mandatory break detection has been factored out
of the default text breaking algorithm to a higher level in order to refactor
the QCharAttributes bitfields and to optimize the implementation for the common case.
Change-Id: Ieb365623ae954430f1c8b2dfcd65c82973143eec
Reviewed-by: Lars Knoll <lars.knoll@digia.com>
|
|
|
|
|
|
|
|
|
|
| |
The code is not handling formats like "HHmm ss" correctly, so it needs
to be documented until such support is provided.
Task-number: QTBUG-26067 QTBUG-26596
Change-Id: Ia456d8020e3e0aa9422e6e6987ac984f308facf9
Reviewed-by: Qt Doc Bot <qt_docbot@qt-project.org>
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
|
|
|
|
|
|
|
|
|
| |
This allows, among other things, to use QVarLengthArray as the target of
a std::back_insert_iterator.
Change-Id: I507f612a23da854bf865780aa0a7e6312f4a896b
Reviewed-by: João Abecasis <joao.abecasis@nokia.com>
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
|
|
|
|
|
|
|
|
|
| |
Replace them with std::upper_bound; this allows for deprecation of
qUpperBound.
Change-Id: Idef01d2228b9a70eee3d52931d7aedb5bb6ba902
Reviewed-by: Marc Mutz <marc.mutz@kdab.com>
Reviewed-by: Robin Burchell <robin+qt@viroteck.net>
|
|
|
|
|
|
|
|
|
| |
Replace them with std::lower_bound; this allows for deprecation of
qLowerBound.
Change-Id: I536e7338eb85ea6c7c1a5bf23121292767927e0b
Reviewed-by: Marc Mutz <marc.mutz@kdab.com>
Reviewed-by: Robin Burchell <robin+qt@viroteck.net>
|
|
|
|
|
| |
Change-Id: Iac5df9edbe16ec841f783bc5266091a6e34e9cab
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
|
|
|
|
|
|
|
|
|
| |
Make C++ class constructors that can be used with only one
required argument 'explicit' to minimize wrong use of the class.
Change-Id: I12ad5b6eb1794108c6b7464a2573e84068733b03
Reviewed-by: Sergio Ahumada <sergio.ahumada@nokia.com>
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
|
|
|
|
|
|
|
|
| |
Add missing external resource link. Provide constructor documentation.
Change-Id: I6d28a19da5c409dd0e0b48e0f693837b63de38f0
Reviewed-by: Qt Doc Bot <qt_docbot@qt-project.org>
Reviewed-by: Jerome Pasion <jerome.pasion@nokia.com>
|