The use of a unique_ptr member requires that the destructor be
out-of-line, since the payload is only forward-declared in the header
file. This is good hygiene, so do it for CompositionWidget, too.
Add 'explicit' and missing = nullptr to both constructors as a
drive-by.
Change-Id: Ied1c89864f90d3f2c13fb4e9a8bbbe2e6fd6f1d7
Reviewed-by: Giuseppe D'Angelo <giuseppe.dangelo@kdab.com>
We have a problem. Our types don't play well with the std unordered
containers, because they do not specialize std::hash. We therefore
force our users to come up with an implementation, hindering
interoperability, since any two developers are unlikely to come up
with compatible implementations. So combining libraries written by
different developers will result in ODR violations.
Now that we depend on C++11, and thus the presence of std::hash, we
still face the problem that the standard does not provide us with a
means to compose new hash functions out of old ones. In particular, we
cannot, yet, depend on C++17's std::hash<std::string_view> to
implement std::hash<QByteArray>, say. There's also no std::hash for
std::tuple, which would allow easy composition by using std::tie().
So piggy-back on the work we have done over the years on qHash()
functions, and implement the std::hash specializations for Qt types
using the existing qHash() functions, with a twist: The standard
allows implementations to provide means against predictable hash
values. Qt has this, too, but the seed is managed by the container and
passed to the qHash() function as a separate argument. The standard
does not have this explicit seed, so any protection must be implicit
in the normal use of std::hash.
To reap whatever protection that std library has on offer, if any, we
calculate a seed value by hashing int(0). This will be subject to
constant folding if there's no actual seed, but will produce a value
dependent on the seed if there is one.
Add some tests.
A question that remains is how to document the specialization. Can we
have a \stdhashable QDoc macro that does everything for us?
Task-number: QTBUG-33428
Change-Id: Idfe775f1661f8489587353c4b148d76611ac76f3
Reviewed-by: Mårten Nordheim <marten.nordheim@qt.io>
Reviewed-by: Qt CI Bot <qt_ci_bot@qt-project.org>
It's the same operations as boost::hash_combine, but without
explicit magic numbers.
Change-Id: Id65b6f6b7be3070e4ec349ed0cc062af473a95b3
Reviewed-by: Giuseppe D'Angelo <giuseppe.dangelo@kdab.com>
... by replacing them with C++11 range-for loops.
To avoid detaches of these mutable Qt containers,
wrap the container in qAsConst().
Change-Id: I9b5c23c723be024b314a7ad873e21ad63b44b348
Reviewed-by: Mårten Nordheim <marten.nordheim@qt.io>
In a bootstrapped build QCAP does not inherit from QObjectPrivate,
meaning it won't get a polymorphic destructor, causing compile errors.
This is necessary in preparation to add virtuals to QCAP.
The "virtual" keyword is protected to avoid further warnings by
compilers or linters telling that it's unnecessary because the base
class' dtor is already virtual.
Change-Id: Ifbd218154f74b4f1d2f2c5a3ef1fc31970f1c24b
Reviewed-by: David Faure <david.faure@kdab.com>
Reviewed-by: Mårten Nordheim <marten.nordheim@qt.io>
0.3.3 plus a few more patches, including the one that adds
MD_BLOCK_CODE_DETAIL::fence_char, which will enable smarter rewriting
of code blocks (indented vs. fenced).
Change-Id: Ibc892369947a8a8edfa4bc20b1df98a5c8153141
Reviewed-by: Gatis Paeglis <gatis.paeglis@qt.io>
If any non-breakable content (such as a link) already went past
80 columns, or if a word ended on column 80, it didn't wrap the rest of
the paragraph following.
Change-Id: I27dc0474f18892c34ee2514ea6d5070dae29424f
Reviewed-by: Gatis Paeglis <gatis.paeglis@qt.io>
When reading a document like
# heading
- list item
and then re-writing it, it turned into
# heading
- # list item
because QTextCursor::insertList() simply calls QTextCursor::insertBlock(), thus
inheriting block format from the previous block, without an opportunity to
explicitly define the block format. So be more consistent: use
QTextMarkdownImporter::insertBlock() for blocks inside list items too. Now it
fully defines blockFormat first, then inserts the block, and then adds it to
the current list only when the "paragraph" is actually the list item's text
(but not when it's a continuation paragraph). Also, be prepared for applying
and removing block markers to arbitrary blocks, just in case (they might be
useful for block quotes, for example).
Change-Id: I391820af9b65e75abce12abab45d2477c49c86ac
Reviewed-by: Gatis Paeglis <gatis.paeglis@qt.io>
Add QAbstractTextDocumentLayout::markerAt(pos) for hit testing.
(Qt Quick TextEdit needs it too.) Finds out whether the position
corresponds to a marker on a paragraph. I.e. it finds checkboxes in
GitHub-flavored markdown. This enables editor classes to toggle
checkboxes by clicking them.
Use it in QTextEdit to add the checkbox toggling feature. Also show the
"pointing finger" cursor when hovering a toggleable checkbox.
Change-Id: I036c967ab45e14c836272eac2cc7c7d652543c89
Reviewed-by: Gatis Paeglis <gatis.paeglis@qt.io>
Solve many duplication between _p_p.h and .cpp, sorted by module and
header name, and move headers for each features at the end of includes.
Change-Id: Ice9189c10d062bc473258b7ace279df057102167
Reviewed-by: David Faure <david.faure@kdab.com>
Reviewed-by: Tor Arne Vestbø <tor.arne.vestbo@qt.io>
By using NSDMI.
Change-Id: I171133b07ba2c7050e0d279caff0c393e03df182
Reviewed-by: Marc Mutz <marc.mutz@kdab.com>
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
The "copy" name led me to mis-believe that it was a workaround
to the class being non-copiable, and was missing copying a
data member.
This is instead deliberate; so rename the function to
"copyContextFrom", documenting that version won't be copied.
The function is still private API, so actually make it private;
QMessageLogContext already befriended the only user.
Finally, make it return *this (so that it can be chained,
if necessary) and noexcept.
Change-Id: I3deb3c8edc863a88ac0c37467b144ec2e20db5ca
Reviewed-by: Marc Mutz <marc.mutz@kdab.com>
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
As borders are always painted for single cells any rounding errors that
occur when mapping from fixed point numbers to integers during raster
painting will cause ugly drawing artifacts like extra pixels between
cells or borders that overlap.
To avoid this the following values are being rounded:
- table offset
- column positions and widths
- row positions and cell heights
For the column positions and widths the distribution constraints (total
table width) are respected.
Task-number: QTBUG-43589
Change-Id: I6308c113f8f9be3db8e9267629f26b5d65d2d7c7
Reviewed-by: Shawn Rutledge <shawn.rutledge@qt.io>
All calls to QINSTALL or QINSTALL_PROGRAM were written into Makefiles
with a preceding dash which lets make ignore the command's exit code.
Scripts (and users) calling 'make install' had no way to determine if an
installation completely succeeded, unless they inspected make's output
or the installation tree.
Remove the leading dash from those commands to stop 'make install' on
failure. Users who really want to ignore the exit codes can run 'make -k
install'.
[ChangeLog][qmake] Installation targets do not ignore the exit code of
QINSTALL and QINSTALL_PROGRAM anymore.
Fixes: QTBUG-18870
Change-Id: I7c072c896e6bd2b2ddba4b9c082c5bf627c90f50
Reviewed-by: Lars Knoll <lars.knoll@qt.io>
Reviewed-by: Simon Hausmann <simon.hausmann@qt.io>
Debug/release install targets can potentially install the same files which
leads to errors on install when running make with -j > 1.
If build_all is set, make the 'install' dependent of the new target
'debug-release-install' which contains the commands of 'debug-install' and
'release-install'.
Of course, debug/release is not hard-coded, but the content of the BUILDS
variable is used.
Change-Id: I67b504a95b83daf43bc89dcc0e3391b67e19c027
Reviewed-by: Christian Kandeler <christian.kandeler@qt.io>
Reviewed-by: Simon Hausmann <simon.hausmann@qt.io>
The qmake code to read output from dependency-generation was adding
QByteArray values to a QString, thereby tacitly converting from UTF-8;
this is misguided. Hopefully, the command emits its output in the same
local 8-bit encoding that QString knows to convert from.
Simplified needlessly verbose loops (that violated Qt coding style) in
the process.
Fixes: QTBUG-75904
Change-Id: I27cf81ffcb63ebc999b8e4fc57abdb9a68c4d2b3
Reviewed-by: Jörg Bornemann <joerg.bornemann@qt.io>
The state of the module is done, not deprecated, so we shouldn't
recommend users to move away from it in all cases. There's also
no direct replacement for the DOM API.
Fixes: QTBUG-70629
Change-Id: Ifaff9757234bd68a411a3da1403c57bbbcb94693
Reviewed-by: Lars Knoll <lars.knoll@qt.io>
Reviewed-by: Leena Miettinen <riitta-leena.miettinen@qt.io>
Reviewed-by: Alex Blasche <alexander.blasche@qt.io>
TSAN does not understand the futex system call.
Now, in QMutex and QSemaphore usage, futex is always wrapped by
atomic operations that always do an acquire (before waiting) and
a release (before waking waiters). That alone realizes a
synchronizes-with, and since Qt uses std::atomics, TSAN knows
what's going on and does not complain.
But what if one uses futex directly, or we change the
algorithms, or introduce some other new synchronization
primitive somewhere? Luckily TSAN offers annotations for this
that we can use.
This patch annotates the main entry point for the futex syscall
with a pair of acquire/release semantics. A futex call
guarantees total ordering on the operations on the futex
word(s), whether the call succeeds or fails.
Change-Id: Ib80ff898c09fbb6fc73989247eb757bf70971a8a
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
This came about trying to remove the Java-style iterator. It was used to
iterate in reverse order, something QMap can't do, easily, due to lack
of rbegin()/rend(). Instead of writing ugly loops, use a vector of pairs,
fill it, sort it, then iterate over that one in reverse.
Change-Id: I09c8a2732a0699fff4c497778745523e20d348a1
Reviewed-by: Paul Wicking <paul.wicking@qt.io>
The noexcept part should be clear. The constexpr should make the
compiler not create initialization code for static instances. This is
how std::mutex works[1], and Clang and GCC understand it. MSVC
doesn't.
[1] https://en.cppreference.com/w/cpp/thread/mutex/mutex
Change-Id: If9e4c6c0279f0024a659d5602d55e126224ca950
Reviewed-by: Giuseppe D'Angelo <giuseppe.dangelo@kdab.com>
Fixing this in Qt 5 would be BIC. Although it could be okayish, there
are many more classes that shouldn't be copyable. I need to go through
clazy's output and fix them, and would rather do this noise for Qt 6 and
leave Qt 5 alone and purely BIC free.
Added a move-ctor and move-assign, as well as swap(), deprecated
copy-ctor and copy-assign.
The new copy special member functions warn at runtime if they are called.
In order to not pollute client code with the warning strings, lock them
away by defining the functions out-of-line.
[ChangeLog][QtCore][QBasicTimer] QBasicTimer is now a move-only class.
Copying is now deprecated and will be removed in Qt 6.
[ChangeLog][QtCore][QBasicTimer] Added swap() member and free function.
Change-Id: Ic3e6a26f3989d4c8d125c06e8b0b825411c6e106
Done-with: Marc Mutz <marc.mutz@kdab.com>
Reviewed-by: Giuseppe D'Angelo <giuseppe.dangelo@kdab.com>
Silence a warnings flood about qVariantFromValue(), QString::null
and QWeakPointer::data().
Change-Id: I73347190c0fa396b39b9efd00447cf24e48259a0
Reviewed-by: Volker Hilsheimer <volker.hilsheimer@qt.io>
It seems like an optimization on the itemsDirty flag
caused a bug to be re-introduced. When a popup is shown
on a new screen, the itemsDirty must however be set to
ensure that new correct sizes are calculated.
Task-number: QTBUG-59794
Change-Id: Ifb5c233b1f9d4d38bd0cd7a9a71cc32ad3212f8c
Reviewed-by: Morten Kristensen <msk@nullpointer.dk>
Reviewed-by: Friedemann Kleint <Friedemann.Kleint@qt.io>
Reviewed-by: Christian Ehrlicher <ch.ehrlicher@gmx.de>
this also optimizes network post method handling
Task-number: QTBUG-75660
Change-Id: Ibb0d01f2cc2b2bc7802598c4f6f04b04882c12ca
Reviewed-by: Morten Johan Sørvig <morten.sorvig@qt.io>
When runing the scenegraph example sgengine, gets rid if this error:
QSGContext::initialize: stencil buffer support missing, expect
rendering errors
Change-Id: I7f3a82409bc2cf81cf8217876e527f9c45be3bf4
Fixes: QTBUG-74694
Reviewed-by: Morten Johan Sørvig <morten.sorvig@qt.io>
This user of QList depended on the stability-of-reference guarantee of QList.
Make that explicit by using a vector of unique_ptrs. Adapt to new API.
This patch does not intend to fix any pre-existing problems with the code,
such as double-lookups. It is focused on getting rid of this questionable
use of QList, so the code doesn't explode when QList temporarily becomes
QVector in wip/qt6. Not more, not less.
Change-Id: I33847f33aa9f411ad9cd6c046653b7ab22a733cb
Reviewed-by: Lars Knoll <lars.knoll@qt.io>
The function is const, but the d-pointer doesn't propagate it (raw pointer),
so we need the qAsConst() even here.
Change-Id: I9d2e1f7715abb3dc67a87cdadaa7ded971b15848
Reviewed-by: Mårten Nordheim <marten.nordheim@qt.io>
The old code iterated over the peer list, inserting active peers into
a QSet, curiously not stopping to search for clients if it had already
determined the peer to be active.
It then iterated over the peers again, storing the indexes of the
peers it had determined to be active in the first loop, in a QList.
It _then_ iterated over the index list, in reverse, calling removeAt()
on the peers list.
<sean parent>That's a remove_if!</sean parent>
The twist is, that only some maximum number of inactive peers should
be removed, just enough to bring the number of peers below a
predefined number.
To solve, use a lambda that keeps track of the number of times it has
returned true, returning false once the count drops to zero. We can't
use a mutable lambda here, since the STL algorithms are allowed to
copy the predicate as many times as they wish, and, indeed, remove_if
is commonly implemented by calling find_if. But the standard
guarantees exactly one application of the predicate per element, so we
can assume that we're not called again on the same element, and
therefore keep a reference to an external count.
With this, what was a horrible mess becomes a single call to remove_if.
Also change a while(--n) c.removeFirst() loop to a single call of
range-erase.
Change-Id: I6c6a54a1805e5b376800e1116e7aec643e95e4e1
Reviewed-by: Mårten Nordheim <marten.nordheim@qt.io>
Fix warning:
src/corelib/kernel/qmetatype.h:1723:27: warning: ‘T* QWeakPointer<T>::data() const [with T = QFile]’ is deprecated: Use toStrongRef() instead, and data() on the returned QSharedPointer [-Wdeprecated-declarations]
shown when compiling tst_QVariant by using
QWeakPointer::internalData().
Change-Id: I5ea543019b4f8e5dfc829939cd2011ae65f12876
Reviewed-by: Marc Mutz <marc.mutz@kdab.com>
Java-iterators are going to be deprecated.
Change-Id: I2e6353f3fd9e2ddaf0767e7f6cea713249d9591e
Reviewed-by: Mårten Nordheim <marten.nordheim@qt.io>
Reviewed-by: Paul Wicking <paul.wicking@qt.io>
QDate looked up lengths of months in an array. Change it to use some
simple arithmetic instead. Benchmark result unchanged:
RESULT : tst_QDate::monthLengths():
0.33 msecs per iteration (total: 87, iterations: 256)
Change-Id: I1ab0fa3b97e6716598e50d643a498e0b01c04a96
Reviewed-by: Volker Hilsheimer <volker.hilsheimer@qt.io>
The two perform related calculations, so test them together.
RESULT : tst_QDate::monthLengths():
0.33 msecs per iteration (total: 87, iterations: 256)
Change-Id: I86b36a9182b00c0a1f40e858ed3e7812434974c4
Reviewed-by: Volker Hilsheimer <volker.hilsheimer@qt.io>
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
When parsing the CLDR data, we only handle language, script and
territory (which we call country) codes if they are known to our
enumdata.py tables. When reporting the rest as unknown, in the
content of an actual locale definition (not the likely subtag data),
check whether en.xml can resolve the code for us; if it can, report
the full name it provides, as a hint to whoever's running the script
that an update to enumdata.py may be in order.
Change-Id: I9ca1d6922a91d45bc436f4b622e5557261897d7f
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
Reviewed-by: Lars Knoll <lars.knoll@qt.io>
Reviewed-by: Konstantin Ritt <ritt.ks@gmail.com>
It was misnamed local_database, quite missing the point of its name.
Change-Id: I73a4fdf24f53daac12304de1f443636d89afacb2
Reviewed-by: Lars Knoll <lars.knoll@qt.io>
Reviewed-by: Konstantin Ritt <ritt.ks@gmail.com>
