In a previous attempt to solve the problem of selection offer
globals arriving too early, the handling was delayed. This
solved the issue of crashing but introduced a timing issue,
because the offers (the mime types) will arrive immediately
after the global and therefore will simply be ignored in
case the delayed processing of the selection offer had not
yet been done. The visibility of the problem depended on the
implementation of the compositor, with recent changes to
qt-compositor the issue is very visible.
The patch solves the issue properly: The wayland clipboard
instance is created right away, as early when needed, and the
integration will simply pick up the already created instance.
Change-Id: I75aaba4b0590c05cc0091bed7bb3593186c1188f
Reviewed-on: http://codereview.qt.nokia.com/687
Reviewed-by: Qt Sanity Bot <qt_sanity_bot@ovi.com>
Reviewed-by: Paul Olav Tvete <paul.tvete@nokia.com>
Remove the references to demos from configure and qtbase.pro.
This is done because of the merge of the demos and examples in Qt.
Everything will be added as an example.
Change-Id: Iec040f5c719384c7aabba971316de40195ed3a69
Reviewed-on: http://codereview.qt.nokia.com/619
Reviewed-by: Qt Sanity Bot <qt_sanity_bot@ovi.com>
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@nokia.com>
OpenSSL versions 0.9.8 and 1.0.0 produce slightly different output when
dumping a certificate.
Change-Id: I2cf27213237a2e1e08f1b0345c29ca2cd441f41c
Reviewed-on: http://codereview.qt.nokia.com/555
Reviewed-by: Qt Sanity Bot <qt_sanity_bot@ovi.com>
Reviewed-by: Martin Petersson <Martin.Petersson@nokia.com>
Adds support for binding "dual stack" sockets (via QUdpSocket or
QTcpServer). A dual stack socket will accept incoming connections on
either IPv4 or IPv6 interfaces.
QHostAddress::Any - use this to bind a dual stack socket
QHostAddress::AnyIPv6 - use this to bind a socket for IPv6 only
QHostAddress::AnyIPv4 - use this to bind a socket for IPv4 only
Binding to a specific address rather than one of the "any" addresses
is restricting you to a protocol anyway so no behaviour change there.
IPv6 sockets were previously dual stack on some OS and v6 only on others
Any previously meant IPv4 only
This commit implemented & tested on Windows 7, Linux (Ubuntu 10.04)
and Mac OS 10.6.7.
Windows XP and server 2003 do not support dual stack sockets, even though
they can support IPv6. On those versions, QHostAddress::Any will still
bind to IPv4 0.0.0.0 (which is also the behaviour anywhere QT_NO_IPV6 is
defined)
Autotests run:
qudpsocket (includes a new test case)
qtcpserver (includes a new test case)
qtcpsocket
qnetworkreply
qhostaddress
Task-number: QTBUG-17080
Change-Id: Id486677c4f832e18dc0ff1a86c5f5fc422c9eb4f
Reviewed-on: http://codereview.qt.nokia.com/421
Reviewed-by: Qt Sanity Bot <qt_sanity_bot@ovi.com>
Reviewed-by: Thiago Macieira <thiago.macieira@nokia.com>
Reviewed-by: Markus Goetz
apparently there are really big certificates around, so 4k might not
always be enough.
Change-Id: I84df82d117469a14b4c6db81e0ceecc1a8ba47b3
Reviewed-on: http://codereview.qt.nokia.com/554
Reviewed-by: Qt Sanity Bot <qt_sanity_bot@ovi.com>
Reviewed-by: Martin Petersson <Martin.Petersson@nokia.com>
We should do FcConfigSubstitute(0, pattern, FcMatchFont) on a
FcPattern for query because the family list in it will contain
almost all the families after FcConfigSubstitute(0, pattern,
FcMatchPattern), then the test in <match target="font"> will
almost always succeed. In general, FcMatchFont substitute
should only be done on the FcPattern that we got after
FcFontMatch() or FcFontRenderPrepare().
Based on the suggestion of fontconfig author Behdad Esfahbod,
this patch reorganized the tryPatternLoad logic so that it only
does the QFontEngine creation part, FcPattern *match is retrieved
outside of that function. In this way, the match pattern we got
can be either from FcFontMatch() or after FcFontRenderPrepare()
on one of the fonts we got from qt_fontSetForPattern(). Then we
don't need to duplicate the pattern and add all criterias back
with qt_addPatternProps(). It not only simplified the code a lot
but also fix the way we apply FcMatchFont substitution. This
substitution will either be done by FcFontMatch() or
FcFontRenderPrepare, both implicitly.
Task-number: QTBUG-2148, QTBUG-19947
Change-Id: I2e4dbbe7f448e81850176732b4a65028f2b912fa
Reviewed-by: Eskil
Reviewed-on: http://codereview.qt.nokia.com/579
Reviewed-by: Qt Sanity Bot <qt_sanity_bot@ovi.com>
Reviewed-by: Jiang Jiang <jiang.jiang@nokia.com>
cb760eaef631abd49836ae5c8dc12a61ef5eff0d changed the way we generate
HTML for empty blocks. tst_QTextDocument::toHtml has to be updated
accordingly.
Reviewed-by: Samuel
(cherry picked from commit 2701802511d9c09a11212cc37838154245b0c0ca)
Change-Id: I0664557f9d74e1aacd46cfcf4cfa2cde4f21a719
Reviewed-on: http://codereview.qt.nokia.com/581
Reviewed-by: Jiang Jiang <jiang.jiang@nokia.com>
Reviewed-by: Qt Sanity Bot <qt_sanity_bot@ovi.com>
All of the benchmarks explicitly turn on release mode. This fails on
Mac and Windows if Qt is configured as debug-only. It is also possibly
misleading on Linux, as compiling the benchmark in release mode does
not imply that Qt is compiled in release mode.
The results are generally not useful if release mode Qt is not
available, so disable them by default in this case.
Change-Id: Ifa2bb6ab6412dd360ff9dfb890ace2f51e0eef86
Reviewed-on: http://codereview.qt.nokia.com/549
Reviewed-by: Qt Sanity Bot <qt_sanity_bot@ovi.com>
Reviewed-by: Jo Asplin <jo.asplin@nokia.com>
Adds a function that will convert a certificate to human readable text
format using the openssl print function. This is useful for debugging
and for displaying the full details of a certificate (including those
parts not supported by the Qt API).
Change-Id: I27238d05df37f8b15ad09f8e761b06344631a9ce
Merge-request: 2
Reviewed-by: Peter Hartmann <peter.hartmann@nokia.com>
Reviewed-on: http://codereview.qt.nokia.com/551
Reviewed-by: Qt Sanity Bot <qt_sanity_bot@ovi.com>
Loads one of the test certs then compares the result with a known good
text version.
Change-Id: I3a0d6a7f60ce0f48e0cd8032c9964e9c4217dca7
Merge-request: 2
Reviewed-by: Peter Hartmann <peter.hartmann@nokia.com>
Reviewed-on: http://codereview.qt.nokia.com/552
Several autotests were assuming that Q_OS_MAC == Mac cocoa port, which
caused compile failures when QPA is used.
Change-Id: I4480ed815c15b6d9ce83edf0057b7293f2e3ad7e
Reviewed-on: http://codereview.qt.nokia.com/533
Reviewed-by: Qt Sanity Bot <qt_sanity_bot@ovi.com>
Reviewed-by: Morten Johan Sørvig <morten.sorvig@nokia.com>
This test was forcing on release mode, which does not work if the user
requested to build Qt in debug mode only, and this affects the library
name (as on Mac and Windows).
Change-Id: I11111c0ffee11111111111111111111111111111
Reviewed-on: http://codereview.qt.nokia.com/532
Reviewed-by: Jo Asplin <jo.asplin@nokia.com>
-Wl,-rpath-link is expected to work for all gcc except mac.
Change-Id: I4b57e0088db49a8e35840d981e5d6edbb1012396
Reviewed-on: http://codereview.qt.nokia.com/506
Reviewed-by: Qt Sanity Bot <qt_sanity_bot@ovi.com>
Reviewed-by: Rohan McGovern <rohan.mcgovern@nokia.com>
This fix is for qtbase libraries with -qtnamespace and -qtlibinfix
options in configure.
Task-number: QTBUG-19964
Change-Id: I2f2ff1748f2c1c2b20d5f73b6be36f68a7a26cef
Reviewed-on: http://codereview.qt.nokia.com/505
Reviewed-by: Qt Sanity Bot <qt_sanity_bot@ovi.com>
Reviewed-by: Rohan McGovern <rohan.mcgovern@nokia.com>
Reviewed-by: Gunnar Sletta <gunnar.sletta@nokia.com>
This redoes f541c78e1bc5b293466b40e6f10496199a4a5d73 in a way which
should be more compliant with different browsers. In particular,
Outlook didn't support the CSS trick in the last fix, so we need the
somewhat larger patch which adds an extra line break node to the tree
and then ignores it when re-importing the document.
Task-number: QTBUG-3669
Reviewed-by: Simon Hausmann
(cherry picked from commit cb760eaef631abd49836ae5c8dc12a61ef5eff0d)
Change-Id: Ia55bf39d52461aa9445a9a5d0bfb5b4c5bf1e9bd
Reviewed-on: http://codereview.qt.nokia.com/492
Reviewed-by: Eskil Abrahamsen Blomfeldt <eskil.abrahamsen-blomfeldt@nokia.com>
Reviewed-by: Qt Sanity Bot <qt_sanity_bot@ovi.com>
On X11 QProcess would emit two started signals when calling
QProcess::waitForStarted(). We should expect that the private
implementation of waitForStarted() should emit the started signal
and return true or false appropriately.
Task-number: QTBUG-7039
Change-Id: I3d381399ab7a39bf57db03a110fa6747a4fc6a24
Reviewed-by: pending
Reviewed-on: http://codereview.qt.nokia.com/331
Reviewed-by: Thiago Macieira <thiago.macieira@nokia.com>
We would include too much in the characters displayed, since
itemEnd is actually the start of the next item and not the
end of the current one. Adding a minus one removes the duplicated
characters at the end of text lines.
Change-Id: I1e13470548fafaa692ae58e019e9e2469a947f5e
Reviewed-on: http://codereview.qt.nokia.com/495
Reviewed-by: Qt Sanity Bot <qt_sanity_bot@ovi.com>
Reviewed-by: Gunnar Sletta <gunnar.sletta@nokia.com>
Adding the fallback to QPainterPath for large QStaticTexts created a
regression where all text would be painted in black regardless of the
color you set on it. The color in QStaticTextItem is sometimes invalid,
in which case the current painter color should be used. In either case,
the color is already set on the current pen when entering
drawStaticTextItem() so we can just use that.
Task-number: QTBUG-19950
Reviewed-by: Jiang Jiang
(cherry picked from commit ee77ee5c25f58271e6f2863225d08573da86c3ee)
Change-Id: I955aa6526e5b14589430f8dccd006a9de9ae08c9
Reviewed-on: http://codereview.qt.nokia.com/491
Reviewed-by: Eskil Abrahamsen Blomfeldt <eskil.abrahamsen-blomfeldt@nokia.com>
cycle_p.h is only needed when HAVE_TICK_COUNTER is defined, and should
not be included otherwise. It is also a 3rd party header which is not
exported as a private header so far, which makes compilation of
dependenty modules fail.
Change-Id: I5c2546eae7d65fc68b8411c20634aca541c327d6
Reviewed-on: http://codereview.qt.nokia.com/465
Reviewed-by: Qt Sanity Bot <qt_sanity_bot@ovi.com>
Reviewed-by: Marius Storm-Olsen <marius.storm-olsen@nokia.com>
When the build directory is the install directory, separate modules
should use qtbase as the path for libs, module pris etc.
This fixes the logic in default_pre.prf so that syncqt is called
correctly in these cases.
Change-Id: Ib3798bfd145d3a3e43b9a222af46270245e6c2e6
Reviewed-on: http://codereview.qt.nokia.com/464
Reviewed-by: Qt Sanity Bot <qt_sanity_bot@ovi.com>
Reviewed-by: Marius Storm-Olsen <marius.storm-olsen@nokia.com>
Reviewed-by: Eckhart Koppen <eckhart.koppen@nokia.com>
Handle selection offer globals properly even if they arrive during the
initial blocking read. If such a global arrives so early, the platform
integration is not yet available from QApplicationPrivate (as it is
just being constructed). Therefore the handling of the selection
offer has to be delayed. At the moment selection offers are typically
posted as globals and not added to the global list, but that is likely
to change in the future.
Change-Id: Ib4ae804ad7f19e05978ee08828b88e028a3bf7b2
Reviewed-on: http://codereview.qt.nokia.com/446
Reviewed-by: Qt Sanity Bot <qt_sanity_bot@ovi.com>
Reviewed-by: Jørgen Lind <jorgen.lind@nokia.com>
which leads to absurd statement (QBuffer::open() == !QBuffer::isOpen()) to be true.
also treat Truncate as (Truncate | WriteOnly) to satisfy lazy ones
Reviewed-by: Joao
Merge-request: 2612
(cherry picked from commit 6b91affb9a355e668bc9d06dee580d95230ac63a)
Change-Id: I657d4d0a33f7993313fe2a1a8ba408371991717f
Reviewed-on: http://codereview.qt.nokia.com/447
Reviewed-by: Qt Sanity Bot <qt_sanity_bot@ovi.com>
Reviewed-by: Peter Hartmann <peter.hartmann@nokia.com>
When QTextDocument exports HTML, it makes an effort to be compatible
with its own importer, hence it has to be compatible with the dialect
of HTML which Qt has developed over the years. One incorrect
interpretation in Qt is that an empty paragraph is interpreted as an
empty line. So if you use a QTextDocument to produce HTML for text where
an empty line has been added, this empty line will not be visible when
the document is viewed in a compliant browser. The fix is to set the
height of the empty paragraph to 1em, so that it will match the current
pixel size of the font, thus look the same as a <p><br /></p> but
without altering the structure of the document.
Reviewed-by: Gunnar
(cherry picked from commit f541c78e1bc5b293466b40e6f10496199a4a5d73)
Change-Id: Ic0eae2c81609b8872eb2eb9344a3ec416cd09149
Reviewed-on: http://codereview.qt.nokia.com/445
Reviewed-by: Qt Sanity Bot <qt_sanity_bot@ovi.com>
Reviewed-by: Eskil Abrahamsen Blomfeldt <eskil.abrahamsen-blomfeldt@nokia.com>
Fonts like "Helvetica Neue UltraLight" or "Skia Regular Black
Condensed" can't be selected in Qt because either they don't
report correct numeric values for weight/stretch/etc. or these
values are not mapped from QFont enums in a linear way. Thus
we provide a shortcut to select these fonts with PostScript
name or full name without resorting to family name matching in
QFontDatabase (these fonts are not registered in font database
anyway). After this, we can simply use:
QFont font("Helvetica Neue");
font.setStyleName("UltraLight");
to select these fonts. QCoreTextFontEngineMulti matched like
this can be created directly from the CTFontRef instance
instead of creating from the font name, making this process
faster.
The commit also cleaned up the font loading process in Mac
font database a bit, moving the code for family matching into
a separate function.
Add QFontInfo::styleName() and QRawFont::styleName() to access
the resolved style name for a font.
Task-number: QTBUG-19366
Change-Id: Iad07768c02ed06cc8d6b7395dec554384f410506
Reviewed-on: http://codereview.qt.nokia.com/333
Reviewed-by: Qt Sanity Bot <qt_sanity_bot@ovi.com>
Reviewed-by: Jiang Jiang <jiang.jiang@nokia.com>
This is done to remove a compile-time dependency on the harfbuzz
source files inside qtbase/src/3rdparty. These are not accessible
now that QT_SOURCE_TREE is not accessible as a qmake variable
anymore.
With the refactor branch we might solve this differently, but for
now this is how we get svg and declarative to compile.
Change-Id: I5dad23f2ea1f650e2621c1c1fcf39632a3a22ae8
Reviewed-on: http://codereview.qt.nokia.com/378
Reviewed-by: Qt Sanity Bot <qt_sanity_bot@ovi.com>
Reviewed-by: Eskil Abrahamsen Blomfeldt <eskil.abrahamsen-blomfeldt@nokia.com>
homeDirPath requires Qt 3 support, which is not enabled in all
configurations.
Change-Id: I7d51b880574021ef4569a4a2b16f06a7786dbb7d
Reviewed-on: http://codereview.qt.nokia.com/415
Reviewed-by: Qt Sanity Bot <qt_sanity_bot@ovi.com>
Reviewed-by: Pierre Rossi <pierre.rossi@gmail.com>
Not having load(qt_module) first in a .pro file which later
does a load(qt_module_config) will likely overwrite
compiler/linker options in the .pro file.
This can be hard to figure out, and wasn't required before
when we used a global .qmake.cache for all modules, since
it 'preloaded' qmodule.pri file for all .pro files.
Change-Id: I8a9aa3247f938be0b3b13f8c8cc96c18a142ba7b
Reviewed-on: http://codereview.qt.nokia.com/412
Reviewed-by: Qt Sanity Bot <qt_sanity_bot@ovi.com>
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@nokia.com>
enableHarfbuzz() should only be defined on Mac.
Reviewed-by: Eskil
(cherry picked from commit 5ce3fbb67b79c3732fd47b296ef9421398ca520c)
Change-Id: I9ecb5db49478c3f5beb5d41cf99320f0faedce2e
Reviewed-on: http://codereview.qt.nokia.com/375
Reviewed-by: Qt Sanity Bot <qt_sanity_bot@ovi.com>
Reviewed-by: Jiang Jiang <jiang.jiang@nokia.com>