summaryrefslogtreecommitdiffstats
Commit message (Collapse)AuthorAgeFilesLines
* Added digit substitution for date/time on Windows.Denis Dzyubenko2011-03-171-7/+52
| | | | | | | Whenever we are asked to convert a date/time object to string, we should respect the system digit substitution rules. Reviewed-by: Zeno Albisser
* Added short time format support on Windows.Denis Dzyubenko2011-03-171-3/+6
| | | | | | | Since Windows 7 there is a new TIME_NOSECONDS flag that is supposed to be used whenever short time format is requested. Reviewed-by: Zeno Albisser
* Merge remote branch 'earth-team/master' into master-i18nDenis Dzyubenko2011-03-171032-24035/+46417
|\ | | | | | | | | Conflicts: src/corelib/tools/qlocale.cpp
| * QTcpSocket: Fix for disconnect in Unbuffered mode.Markus Goetz2011-03-162-5/+10
| | | | | | | | | | | | | | | | We need to set the state properly after we have been disconnected. Also fix the HTTP layer, it needs to handle the disconnect when reading. Reviewed-by: Peter Hartmann
| * Don't cache resolved paths in MakefileGenerator::fileFixify().Andreas Kling2011-03-152-60/+0
| | | | | | | | | | | | | | | | | | | | | | The cache ends up dog-slow because of poor distribution in hashing of QStrings, and even with a better hashing function, computing the result on the fly is still faster. Tested with WebCore.pro, on Linux processing time goes from 58 to 8 seconds, and on Windows from 105 to 65 seconds. Reviewed-by: Marius Storm-Olsen
| * Option::fixString: Don't recreate same QRegExp on each function call.Andreas Kling2011-03-151-1/+1
| | | | | | | | Reviewed-by: Oswald Buddenhagen
| * Option::fixString: Avoid double hash lookups for cache hits.Andreas Kling2011-03-151-4/+6
| | | | | | | | Reviewed-by: Oswald Buddenhagen
| * tst_qhostinfo: FixesMarkus Goetz2011-03-151-1/+2
| | | | | | | | Reviewed-by: Peter Hartmann
| * QNAM HTTP: error() in case connection is closed unexpectedlyMarkus Goetz2011-03-152-8/+31
| | | | | | | | | | | | Task-number: QT-4658 Task-number: QT-3494 Reviewed-by: Peter Hartmann
| * SSL: give protocol enum SecureProtocols an own valuePeter Hartmann2011-03-154-6/+8
| | | | | | | | | | | | | | | | ... so that an application that uses SecureProtocols can make use of updates to a Qt version without being recompiled. Reviewed-by: Markus Goetz Reviewed-by: Richard J. Moore
| * QMAKE: Fix post build events for VS2010Martin Petersson2011-03-151-3/+8
| | | | | | | | | | | | | | | | We have changed 'Use in build' for VS2010 to now use "Excluded From Build" to have the same as in VS2008. This will enable post build event again for VS2010. Reviewed-by: Joerg Bornemann
| * SSL backend: avoid setting SNI hostname for old SSL versionsPeter Hartmann2011-03-142-2/+5
| | | | | | | | | | | | | | | | With this patch, we only use SNI functionality when the SSL version supports it (meaning when using TLS), otherwise the function call would trigger a warning. Reviewed-by: Markus Goetz
| * SSL: Switch default version to TlsV1SslV3 (i.e. use TLS 1 or SSL 3)Peter Hartmann2011-03-146-8/+23
| | | | | | | | | | | | | | | | ... and introduce a new enum SecureProtocols. Switching the default version is better for compatibility (e.g. servers using this option will understand both TLS and SSL 3). Reviewed-by: Markus Goetz
| * SSL: introduce new option TlsV1SslV3 for SSL communicationPeter Hartmann2011-03-144-31/+130
| | | | | | | | | | | | | | | | | | | | currently there are 3 supported protocols: SSL2, SSL3 and TLS1. SSL2 is considered insecure and should not be used anymore. This commit offers an option to use both TLS1 and SSL3, leaving SSL2 out. Part-of-the-patch-by: Darren Lissimore Reviewed-by: Markus Goetz Task-number: QTBUG-12338
| * Mac Style: Compile FixMartin Petersson2011-03-141-1/+1
| | | | | | | | | | Task-number: QTBUG-13055 Reviewed-by: Markus Goetz
| * Mac: Center the Window title on Cocoa.Martin Petersson2011-03-101-0/+2
| | | | | | | | | | Task-number: QTBUG-13593 Reviewed-by: Richard Moe Gustavsen
| * Fix corner of scroll area so it is stylable on Mac.Martin Petersson2011-03-101-1/+1
| | | | | | | | | | Task-number: QTBUG-13055 Reviewed-by: Olivier Goffart
| * QNAM HTTP: Pair channels with requests at a later state.Martin Petersson2011-03-103-24/+48
| | | | | | | | | | | | | | | | | | | | | | Connect a new channel only when there is no one currently available. Only pair the request with the channel once we know if it got connected or not. This leads to faster sending on requests as we reuse already connected channels that became free in the meanwhile. Task-number: QTBUG-17084 Reviewed-by: Markus Goetz Reviewed-by: Peter Hartmann
| * tst_qnetworkreply: fix MiniHttpServer crashMartin Petersson2011-03-101-1/+0
| | | | | | | | Reviewed-by: Markus Goetz
| * Windows: Activate the context menu on tray icons when shown.Martin Petersson2011-03-101-0/+1
| | | | | | | | | | Task-number: QTBUG-14807 Reviewed-by: Markus Goetz
| * SSL backend: check at runtime for the right OpenSSL version for SNIPeter Hartmann2011-03-044-3/+9
| | | | | | | | | | | | | | | | | | | | | | SNI = Server Name Indication. The function "SSL_ctrl()" has been there since always in OpenSSL, but not with the specific enum SSL_CTRL_SET_TLSEXT_HOSTNAME, so let's avoid the call for older versions. Additionally, fix the resolving of SSL_CTX_load_verify_locations for Symbian (is not used in Symbian yet). Reviewed-by: Markus Goetz
| * QNAM HTTP: Fix the ioPostToHttpFromSocket auto testMartin Petersson2011-03-043-7/+18
| | | | | | | | | | | | | | Introduced a ClosingState for the channel to better track the state of the socket. Reviewed-by: Markus Goetz
| * Disable capabilities example for symbian-gcce due to a bug in elf2e32axis2011-03-031-0/+3
| | | | | | | | | | | | | | | | | | | | | | | | | | | | Phonon exports template instantiations. These exports get weak symbol binding, which is correct according to the C++ ABI, but the problem is that elf2e32 has a bug which does not transfer the weak symbols correctly to the dso file. Therefore, the example will work if you have a prebuilt Qt version and use GCCE, but not if you build Qt from scratch using GCCE. For normal non-template symbols it is not a problem since they get global bindings. RVCT also produces global bindings. RevBy: Shane Kearns
| * Fixed documentation for QByteArrayDenis Dzyubenko2011-03-031-16/+19
| | | | | | | | | | | | | | The data returned by constData() will not be null-terminated if the QByteArray was created from an existing raw data with QByteArray::fromRawData(). Reviewed-by: Oswald Buddenhagen
| * Improve handling QByteArray with QStringBuilderDenis Dzyubenko2011-03-033-5/+30
| | | | | | | | | | | | | | | | | | Instead of relying on the old behavior of the deprecated QString(QByteArray) constructor which stops copying the byte array at the null termination character if there is one embedded in the array, we copy the whole provided QByteArray object into the QString when QStringBuilder is used. Reviewed-by: Harald Fernengel
| * Do not handle posted events in QSplashScreen.Denis Dzyubenko2011-03-032-4/+10
| | | | | | | | | | | | | | | | When QSplashScreen waits for the window to be fully shown, we shouldn't re-enter the event loop by processing posted events. Reviewed-by: axis Reviewed-by: Olivier Goffart
| * SSL tests: Be more verbose in on-demand cert testMarkus Goetz2011-03-032-2/+2
| |
| * Fix qstringbuilder test.Olivier Goffart2011-03-021-0/+1
| | | | | | | | | | | | when QT_NO_CAST_FROM_ASCII is not defined, the string is assigned to an string with more than ascii. So if the codecForLocale is not latin1, it would fail.
| * Wrap qPrintable inside QStringOlivier Goffart2011-03-012-1/+3
| | | | | | | | | | | | | | | | | | | | QString() is a no-op if string is already a QString And it fixes the compilation if other types are use that do not have toLocal8Bit such as QStringBuilder. WebKit trunk has an instance of such usage. Reviewed-by: Joao
| * Fix QTextStream::pos() causes buffer offset issues on large text files.Chris Kawano2011-03-012-0/+37
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | When reading large files that are greater in size than the TEXTSTREAM_BUFFERSIZE buffer, calling pos() caused erroneous buffer positioning. The cause of this was QTextStreamPrivate::readBufferOffset value being updated, but QTextStreamPrivate::readConverterSavedStateOffset was not. Usually, if pos() is called, it is called multiple times. Since QTextStreamPrivate::readConverterSavedStateOffset was never reset, QTextStreamPrivate::readBufferOffset would accumulate the wrong position and QTextStreamPrivate::consume() will 'discard' the wrong number of bytes from the buffer. Task-number: QTBUG-9814 Merge-request: 2569 Reviewed-by: Olivier Goffart <olivier.goffart@nokia.com>
| * Don't compile qtconfig on symbianOlivier Goffart2011-02-281-1/+1
| | | | | | | | Reviewed-by: Benjamin Poulain
| * qtconfig: don't connect to qt3support signalsOlivier Goffart2011-02-282-7/+8
| |
| * Fix errors and warnings in qtconfig.Olivier Goffart2011-02-282-2/+2
| | | | | | | | | | | | That occured with the removal of the qt3support dependency. The setIcon call was added after the merge request was created.
| * Remove qt3support dependency of qtconfigOlivier Goffart2011-02-2823-2798/+1496
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | Revert "Revert the integration of the merge request 2286." This reverts commit e3edad83a1e6e99a551d40d4118352435c6dd710. This had been submitted in the Qt 4.7, but reverted because the change was to intrusive to be pushed in a stable branch. Now re-apply the change in master. Conflicts: tools/qtconfig/mainwindowbase.cpp tools/qtconfig/mainwindowbase.h tools/qtconfig/paletteeditoradvancedbase.cpp tools/qtconfig/paletteeditoradvancedbase.h tools/qtconfig/paletteeditoradvancedbase.ui tools/qtconfig/previewwidgetbase.cpp tools/qtconfig/previewwidgetbase.h tools/qtconfig/previewwidgetbase.ui tools/tools.pro
| * SSL TLS extension on Symbian: work around missing symbolPeter Hartmann2011-02-281-0/+6
| | | | | | | | | | | | | | ... by defining it ourselves. That symbol is missing in the header files for Symbian. Reviewed-by: Shane Kearns
| * SSL: Switch default version to TLS 1.0Peter Hartmann2011-02-285-9/+9
| | | | | | | | | | | | | | | | | | | | | | TLS is backward compatible, so servers only supporting SSL 3 should still work. All browsers send a TLS 1.0 Client Hello these days. However, some servers apparently have problems with a TLS handshake (and a SNI message); for now, wait and see how many of them are broken and either add a fallback to SSLv3 or blacklist them (i.e. set the used SSL version for those servers explicitly). Reviewed-by: Markus Goetz
| * QSslSocket backend: resolve symbols for SNI for SymbianPeter Hartmann2011-02-281-0/+3
| | | | | | | | Task-number: QTBUG-1352
| * Add QSslSocket::setPeerVerifyName()/peerVerifyName()David Faure2011-02-282-0/+31
| | | | | | | | | | | | | | | | | | This allows to set the sslPeerName even when not using connectToHostEncrypted, but rather connectToHost + startClientEncryption Task-number: QTBUG-1352 Merge-request: 1110 Reviewed-by: Peter Hartmann <peter.hartmann@nokia.com>
| * QSslSocket SNI: prefer verificationPeerName then peerName then hostNameDavid Faure2011-02-281-2/+5
| | | | | | | | | | | | | | | | | | As suggested by p--hartmann in a comment for MR 1574. Task-number: QTBUG-1352 Merge-request: 1110 Reviewed-by: Peter Hartmann <peter.hartmann@nokia.com>
| * Add Server Name Identification (RFC4366 section 3.1)Daniel Black2011-02-284-0/+22
| | | | | | | | | | | | | | | | ...to client QSslSocket connections when supported by openssl as per task tracker id #188841 Merge-request: 1574 Reviewed-by: Peter Hartmann <peter.hartmann@nokia.com>
| * Merge remote branch 'earth/master' into masterJoão Abecasis2011-02-2853-638/+2109
| |\
| | * Readded a ';;' that was removed by mistake.axis2011-02-281-0/+1
| | | | | | | | | | | | RevBy: Trust me
| | * Fixed mkspec detection for Symbian.axis2011-02-251-3/+2
| | | | | | | | | | | | RevBy: Liang Qi
| | * Removed javascript-jit from default symbian-gcce build.axis2011-02-251-0/+2
| | | | | | | | | | | | | | | | | | It fails the build anyway. RevBy: Liang Qi
| | * QAbstractSocket: Check for socket state on Unbuffered readsMarkus Goetz2011-02-251-0/+2
| | | | | | | | | | | | | | | | | | | | | In the Unbuffered QTcpSocket mode the read() call could go to the socket engine even though we are not connected right now. Reviewed-by: Peter Hartmann
| | * Corrected a mismerge in GCCE link parameters.axis2011-02-241-1/+1
| | | | | | | | | | | | RevBy: Trust me
| | * Merge branch 'master' of scm.dev.nokia.troll.no:qt/qt into earth-masteraxis2011-02-24874-19632/+39788
| | |\ | | | | | | | | | | | | | | | | | | | | Conflicts: src/network/access/qnetworkreplyimpl.cpp tests/auto/moc/tst_moc.cpp
| | * | QAbstractSocket: Check for engine validity on Unbuffered readsMarkus Goetz2011-02-241-2/+5
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | Do not call read() if the socket engine is already invalid. This could happen when using the Unbuffered QTcpSocket mode which is currently only used inside QNetworkAccessManager. Reviewed-by: Shane Kearns
| | * | tst_qnetworkreply: fix the MiniHttpServer.Martin Petersson2011-02-241-1/+5
| | | | | | | | | | | | | | | | Reviewed-by: Markus Goetz
| | * | tst_qnetworkProxyFactory: fix debug output.Martin Petersson2011-02-241-2/+2
| | | | | | | | | | | | | | | | Reviewed-by: Markus Goetz
generated by cgit v0.12 at 2025-09-12 23:37:24 (GMT)