reduce a bit of code dupe while assembling feature search path

instead of having a bunch of nested loops, collect into a temporary list
and process it at the end.

Change-Id: I97e5642f7e13f7c7b69eae00833e61cdf46a02ed
Reviewed-by: Mark Brand <mabrand@mabrand.nl>

unset QT_MODULE_* after loading the module pris

abuses have been observed in the wild, so make sure these variables are
not available.

Change-Id: I502c3f5db7d341cf6a8bd2ec09e87f129da2fca6
Reviewed-by: Mark Brand <mabrand@mabrand.nl>

purge "dumping all variables" debug loops

the as-we-go dump is sufficient (and usually necessary to actually find
the problem). if only the summary is interesting, the -E option can be
used now.

Change-Id: I9e34c6db9dcb99b38013c4d0cb80b8cb88ca36b5
Reviewed-by: Mark Brand <mabrand@mabrand.nl>

repurpose deprecated -E switch

instead of dumping the variables as we are going, dump everything at the
end. this is potentially useful, as opposed to the previous
functionality which was redundant with -d.

Change-Id: Icf14703cb93e03f7079dfc0266b219ad9c902133
Reviewed-by: Mark Brand <mabrand@mabrand.nl>

make a bunch of functions const

values() and variables() get both const and non-const overloads

Change-Id: Idfabea1acc488bf78f24edb831681ee07f0074c4
Reviewed-by: Mark Brand <mabrand@mabrand.nl>

const correctness for variables()

Change-Id: Id11453ccbae3ffa32580badadada945d342d727a
Reviewed-by: Mark Brand <mabrand@mabrand.nl>

don't overuse variables()

Change-Id: Ib2da88e5e29c0bfecc39a9c06e82a34a4bd0b1e6
Reviewed-by: Joerg Bornemann <joerg.bornemann@nokia.com>

make magicValues() const

the weird debugging feature is not used anyway

Change-Id: I07f481a94f2b2ab2a5b61270f0e00183cefd4cd1
Reviewed-by: Joerg Bornemann <joerg.bornemann@nokia.com>

const correctness for values() calls

Change-Id: I1476ae8d3fb7364686398c5650729dee94a8e933
Reviewed-by: Mark Brand <mabrand@mabrand.nl>

don't use backwards compat mapping for generator accesses

we can rely on only user code needing variable mapping, so apply it only
in the evaluator.

Change-Id: I6fc58e7bcf24cf0fa8783d5341ab1e7b9f001c88
Reviewed-by: Joerg Bornemann <joerg.bornemann@nokia.com>

introduce magicValues() function

the only place where the two remaining magic values need to be
referencable is doVariableReplaceExpand(), so make a separate function
and use it only in that place.

Change-Id: I6e2fcfa3a4f16727d90ace56eb88fc99ef272ffc
Reviewed-by: Joerg Bornemann <joerg.bornemann@nokia.com>

make $$PWD non-magic on read

like the other variables, we can just store it in the hash.

Change-Id: I49ad39dca8d498119b27f16ea4bdc44ae698d72e
Reviewed-by: Joerg Bornemann <joerg.bornemann@nokia.com>

set $$_DATE_ only on project startup

this changes the semantics a bit - it will be the datetime of qmake
startup rather than the time a particular file is processed. i'd argue
that this is insignificant.

Change-Id: I75918967bef25038ce54aa81ab03c027384c0268
Reviewed-by: Joerg Bornemann <joerg.bornemann@nokia.com>

deprecate $$IN_PWD

it suggests a symmetry to $$OUT_PWD which simply isn't there. the
shorter alias $$PWD is much more popular anyway.

Change-Id: Iefbfd56f1a3e526f15c9d6cd1bf59778be6f2f8d
Reviewed-by: Mark Brand <mabrand@mabrand.nl>

make project-related variables non-magic

$$_PRO_FILE_, $$_PRO_FILE_PWD_ and $$OUT_PWD can be cleanly initialized.
no need for magic.

Change-Id: I2e339d17bae42ecb573c2f82c716f6be15a35b98
Reviewed-by: Joerg Bornemann <joerg.bornemann@nokia.com>

move initialization of $$TARGET to setupProject()

that's what we have it for

Change-Id: Ib734b97d337a82c73b086dfeb2b263171f2a02b7
Reviewed-by: Joerg Bornemann <joerg.bornemann@nokia.com>

make $$TEMPLATE magic on write instead of on read

Change-Id: I03a68993dd7d4488f8212132a9d492d135295b9e
Reviewed-by: Joerg Bornemann <joerg.bornemann@nokia.com>

make a bunch of invariant variables non-magic

instead of resolving them on-demand, just initialize the value hash
with them. less magic and faster.

Change-Id: I28cb6c21ae6ae60a33734f62acdef0794420ba8f
Reviewed-by: Joerg Bornemann <joerg.bornemann@nokia.com>

don't make $$TEMPLATE_PREFIX available any more

this is a purely internal variable. projects are supposed to
query $$TEMPLATE only.

Change-Id: I32a3aa7012a4fedcf6e77e2e1302ed978baac700
Reviewed-by: Joerg Bornemann <joerg.bornemann@nokia.com>

move initialization of numerous variables to spec_pre.prf

this is cleaner by design and allows removing some hacks.

Change-Id: I3270195b5d62caa476ffde7c1e1ef43cec99c565
Reviewed-by: Joerg Bornemann <joerg.bornemann@nokia.com>
Reviewed-by: Mark Brand <mabrand@mabrand.nl>

remove QMAKE_EXT_H_MOC and QMAKE_CPP_MOD_MOC from Option

these are not actually used inside qmake since ever the code was factored
into moc.prf.

Change-Id: I545f4857ca3f0b4bf2439703700069ac67ad4ca2
Reviewed-by: Joerg Bornemann <joerg.bornemann@nokia.com>

purge QMAKE_MOD_SYSTEM_ENV from Option & co.

this is entirely unused

Change-Id: Ia0368b1d90f61b64178e28479a2088a19aa418e3
Reviewed-by: Joerg Bornemann <joerg.bornemann@nokia.com>

don't read QMAKE_DIRLIST_SEP back into Option::dirlist_sep

it's really only a host setting which is not used any more past the now
removed assignment.

Change-Id: I62c61c893697eb9a7e7be550311bf152d5a8206e
Reviewed-by: Joerg Bornemann <joerg.bornemann@nokia.com>

don't make prf_ext overridable

it's entirely pointless

Change-Id: I1da3c05a0b4f6ac8cea6ecc64ce3a237957440f5
Reviewed-by: Joerg Bornemann <joerg.bornemann@nokia.com>
Reviewed-by: Mark Brand <mabrand@mabrand.nl>

don't copy files to default specs any more

now that the default specs really only forward to the real specs,
it is not necessary to copy any auxiliary files.

Change-Id: I169a61a045063b796062fe6af3a2afbe3f1c9da0
Reviewed-by: Mark Brand <mabrand@mabrand.nl>

don't put QMAKESPEC_ORIGINAL into INCLUDEPATH any more

the spec does that itself, given that the real spec is just included
nowadays, instead of copied (which never worked without side effects).

Change-Id: Ibf655b9a943dadb949d3c7a58d8fe50fcd62cef7
Reviewed-by: Joerg Bornemann <joerg.bornemann@nokia.com>

fix specdir()

get the actual spec path directly from the project - the specs in Option
are not necessarily resolved.

Change-Id: Ia2bf2199c5269aa5b5a9d4c38de36171f25d448b
Reviewed-by: Joerg Bornemann <joerg.bornemann@nokia.com>

use real qmakespec as basis for finding features

doesn't make much of a difference on unix (as the default specs are just
symlinks).
on windows, it makes the gross hack used for finding spec-specific wince
default_post.prfs unnecessary.

Change-Id: Id403dce5be487e1ae22c1f54b8095a6afdd98bc8
Reviewed-by: Joerg Bornemann <joerg.bornemann@nokia.com>

resolve real qmakespec name right after loading it

this is cleaner than resolving it on-demand, as it avoids statics (with
potential side effects on dynamic spec switching).

Change-Id: I2bc15a4c3108376e1b4a01351875fe0c445ee5d5
Reviewed-by: Mark Brand <mabrand@mabrand.nl>

remove support for loading features relative to current directory

it's completely counterproductive. just include() the file instead.
i don't think anyone knew about this "feature", so just removing it.

as a side effect this removes the repeated existence check of already
found feature files, as we can use a clean else-if cascade.

Change-Id: I5d38d38d0a897f2e8857ac68d5649fd4367941c4
Reviewed-by: Joerg Bornemann <joerg.bornemann@nokia.com>

revamp handling of qmake target mode

the project evaluator becomes oblivious of the target mode.
the mode is set up in spec_post.prf according to the spec.
$$QMAKE_TARGET contains the feature suffixes to search, and is also
contained in $$CONFIG.
the target_mode variable itself becomes private to the Makefile class.

Change-Id: I3c06d9dab536b753343cec6c5c491d3203e50bd8
Reviewed-by: Mark Brand <mabrand@mabrand.nl>

read spec_pre and spec_post features

these are read before the qmakespec and before the cache, resp.
this will allow moving some hard-wired logic out of qmake.

Change-Id: I6a63050d7798bc30a4add8c009bcd801a29a0deb
Reviewed-by: Joerg Bornemann <joerg.bornemann@nokia.com>

prune feature search paths which don't actually exist

this should significantly cut down the time wasted looking for files in
non-existing directories, in particular on windows.

Change-Id: I7ab3523fe8c028e3787ebc78e4543ab04f53448e
Reviewed-by: Joerg Bornemann <joerg.bornemann@nokia.com>

eliminate some use of base_vars

this limits references to this variable to its setup and explicit cache
manipulation.

Change-Id: I88dd2418051501abea201f223da7759a15f1c249
Reviewed-by: Joerg Bornemann <joerg.bornemann@nokia.com>
Reviewed-by: Mark Brand <mabrand@mabrand.nl>

purge the integrity target mode

it wasn't really wired, so the only effect it had was disabling the unix
scopes (despite an attempt to override that in the spec, which probably
worked before qt 4.6, and will start working again soon).

Change-Id: I95daff75b508edaf83a8a06fd327350acd62b124
Reviewed-by: Rolland Dudemaine <rolland@ghs.com>
Reviewed-by: Mark Brand <mabrand@mabrand.nl>

don't let the make spec determine the host mode any more

(some of) the generators are (moderately) x-platform, so it makes no
sense to nail the host platform to a generator (and thus a spec).
overriding the host platform is only a debugging mesasure anyway, so one
can use the (now undocumented) -unix/-macx/-win32 options for that.

Change-Id: If2a059f1feeb2c726e5838625ede1c7add829985
Reviewed-by: Joerg Bornemann <joerg.bornemann@nokia.com>
Reviewed-by: Mark Brand <mabrand@mabrand.nl>

repurpose -unix/-macx/-win32 options

they have been deprecated long enough, so we can mess with them now.
don't make them set the target mode any more; the makespec can do that
autoritatively. instead, they can be used to override the host mode for
debugging purposes.

Change-Id: Ife5fdcf4f1b6b926901a80896ac92b7d821278d1
Reviewed-by: Joerg Bornemann <joerg.bornemann@nokia.com>

do not unnecessarily mess with CONFIG

it could/would bite us later

Change-Id: I73f989e7603c6e5b7b85fc4ee4ad2557c1b02d80
Reviewed-by: Mark Brand <mabrand@mabrand.nl>

Add \since 5.0 to QDir::removeRecursively

Change-Id: I2392f26e666d5e01e71932f1afa48a65704f6d48
Reviewed-by: João Abecasis <joao.abecasis@nokia.com>

Remove debug output

Change-Id: I07270211ee6f0439dc7249c8946795a68025828f
Reviewed-by: Konstantin Ritt <ritt.ks@gmail.com>
Reviewed-by: David Faure <faure@kde.org>

Workaround to compile configureapp.cpp with MSVC2010 x64

Remove a non-ascii character from configureapp.cpp to let it to be
compiled successfully with MSVC2010 x64.

Change-Id: I0b016630be49e8731cc438abca4ef959124138be
Reviewed-by: Rohan McGovern <rohan.mcgovern@nokia.com>
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>

Remove all references to XLFD fonts

XLFD fonts are no longer supported in Qt 5.

Change-Id: I83400dab417c933d5cd956c0d168c45b9d79dab7
Reviewed-by: Casper van Donderen <casper.vandonderen@nokia.com>
Reviewed-by: Samuel Rødal <samuel.rodal@nokia.com>

don't compile the host tools for wince already in configure

qmake can now produce proper mixed-target projects

Change-Id: I797f055f6e1487b9aefb75eee91d6c2cc4e6e56e
Reviewed-by: Andreas Holzammer <andreas.holzammer@kdab.com>
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@nokia.com>

Convert QPA menus to use QIcon, and enable icons.

Icon support was disabled in QPA menus, now QIcon is in QtGui, convert the QPA interface, enable setting the icon in QMenu, and make the Cocoa implementation use it. (And fix a ref-counting leak if an icon is actually set)

Change-Id: Ica203bf6826b79d8beee58f39febc851b9633a66
Reviewed-by: Christoph Schleifenbaum <christoph.schleifenbaum@kdab.com>
Reviewed-by: Friedemann Kleint <Friedemann.Kleint@nokia.com>
Reviewed-by: Bradley T. Hughes <bradley.hughes@nokia.com>

Implement RFC6265 test suite

Tests against the test data from the IETF working group
https://github.com/abarth/http-state

The test data is in the parser.json file, imported from that repository
and with one patch applied to make the ordering0001 test case data match
the raw files which are used by their python test server.

Task-number: QTBUG-18920
Change-Id: I17c1a8d92aef2850907f009667c6574e4c8d0cdb
Reviewed-by: Richard J. Moore <rich@kde.org>

QNetworkCookie - ignore unparsable expires attribute

As required by RFC6265, if the date can't be parsed then the
attribute should be ignored but not the whole cookie.

Task-number: QTBUG-15794
Task-number: QTBUG-16798
Change-Id: I2ed2fbbaf2d00b194995349c24b174032a599703
Reviewed-by: Richard J. Moore <rich@kde.org>

Ignore empty domain attribute in cookies

As recommended by RFC6265.
This fixes the optional-domain0042 test case.

Task-number: QTBUG-15794
Change-Id: I6dd459797afcb52fa2a78437f8481f5abc6f3105
Reviewed-by: Richard J. Moore <rich@kde.org>

Use RFC6265 rules for cookie path & path matching

Url encoding of paths is no longer used. This matches the
current release behaviour of Firefox, Chrome and MSIE browsers.
RFC6265 does not allow this type of encoding.

This fixes remaining path test cases in the IETF test suite.
Currently the path0027 test is passed by Firefox but failed by
Chrome and MSIE, so there is a potential compatibility issue.
However it is a corner case with a malformed cookie.

Task-number: QTBUG-15794
Change-Id: I9b02bb5adc32d614f512d314d06f2c60894aa2b0
Reviewed-by: Richard J. Moore <rich@kde.org>

QNetworkCookie: Use RFC6265 rules for parsing Set-Cookie

The ';' separator takes priority even inside a quoted string.
Quotation marks have no special meaning, they are not parsed and
regenerated anymore. This means it is not possible to include
the ';' character inside a cookie value.
Other characters are returned transparently, including [",\]

Task-number: QTBUG-15794
Task-number: QTBUG-26002
Task-number: QTBUG-11641
Change-Id: I4eefef5c6ac7753d5a21c226169e264578521fe9
Reviewed-by: Richard J. Moore <rich@kde.org>

QNetworkCookie - use RFC6265 rules for max-age

If unparsable, ignore the max-age attribute but process the rest of
the cookie normally.
If max age <= 0, set expiration time to "earliest representable time"
To keep this a safe value for conversions, time_t of 0 is used.

This fixes cases 0019 and comma0005 in the test suite.
Due to this change, cookies may be sent after they should have expired
in case the max-age was malformed. Previously they would have been
discarded immediately, which is more likely to break web services.

Task-number: QTBUG-15794
Change-Id: I7882af8eb37db156785e4e358ca639e90c94f8d0
Reviewed-by: Richard J. Moore <rich@kde.org>

QNetworkCookie - ignore unknown attributes

RFC6265 clarifies that unknown cookie attributes should be ignored,
including the version attribute which was defined by RFC2109 but
not used correctly in practice.

This fixes case 0008 in the test suite with minimal risk.

Task-number: QTBUG-15794
Change-Id: I6f15e8e5e2e5f1ed168fc733a5c84d606a452252
Reviewed-by: Richard J. Moore <rich@kde.org>

qget test - fix null pointer dereference

If the server sends a redirect with no body, the file is null

Change-Id: I49fd1d8a4cdd404497ebef4c7f3b478960776896
Reviewed-by: Richard J. Moore <rich@kde.org>

Change invalid test case (two cookies separated by ,)

Two cookies in a single Set-Cookie header are no longer allowed.
Check that this header is parsed according to RFC6265 rules instead

Change-Id: Ice48bbe78a9886208f7d1186cf1d8c37f46f1252
Reviewed-by: Richard J. Moore <rich@kde.org>

QNetworkRequest autotest: adapt to new cookie parsing (re. commas)

Cookies cannot be separated by commas anymore, but are separated by
new lines.
See "Remove support for multiple cookies in one Set-Cookie header to
follow RFC6265."

Reviewed-by: Martin Petersson
Task-number: QTBUG-21456
(cherry-picked from 5d809703aa2d2a08ae7e9610fd42025b081d3d0c)

Change-Id: If7d1b4e58399a5d678495af6ff280409ba220e86
Reviewed-by: Shane Kearns <shane.kearns@accenture.com>

Remove support for multiple cookies in one Set-Cookie header to follow RFC6265.

This also allows cookie values to contain commas to increase compatibility like
most popular browsers do even though the RFC still reserves them for future uses.

Reviewed-by: Peter Hartmann <peter.hartmann@nokia.com>
Task-number: QTBUG-21456
(cherry-picked from 8ba781b01e900148fec2e9d26485369b3295487f)

Change-Id: Ib09ab2411dddf7f99de1c0c31680428b7412fc7e
Reviewed-by: Shane Kearns <shane.kearns@accenture.com>

Add a method for querying window activation status from QPA.

Add QPlatformWindow::isActive(), where the platform
can do further isActive tests, and Windows implementation for it.

Change-Id: I1acfc44d3a4ab36a3aaee52fb7b5f5b40661095e
Reviewed-by: Miikka Heikkinen <miikka.heikkinen@digia.com>
Reviewed-by: Friedemann Kleint <Friedemann.Kleint@nokia.com>
Reviewed-by: Girish Ramakrishnan <girish.1.ramakrishnan@nokia.com>

Fixed missing way of choosing EGL renderable type with QSurfaceFormat.

This has been long overdue, since EGL now lets you choose between
desktop and ES based OpenGL. We also add OpenVG for those who want to
use raw OpenVG with a QOpenGLContext. The underlying EGL API for using
OpenGL / OpenVG is the same, with eglMakeCurrent() and eglSwapBuffers().

Change-Id: Ib0146b3fde5fe632069ebf99e7712f496ee7ea4d
Reviewed-by: Gunnar Sletta <gunnar.sletta@nokia.com>
Reviewed-by: Lars Knoll <lars.knoll@nokia.com>

Remove some unused variables from QAccessibleWidgetPrivate

Change-Id: I22fd67d63131d48356023c26eb0689587ce96967
Reviewed-by: Frederik Gladhorn <frederik.gladhorn@nokia.com>

Fix QCOMPARE for QIcon in QtGui

There's no need of QtWidgets for using QCOMPARE on two QIcons, as
QIcon lives in QtGui.

Change-Id: I40c3d4aeb15fb95876449383d9e2dd1ad39aa5f9
Reviewed-by: Konstantin Ritt <ritt.ks@gmail.com>
Reviewed-by: Olivier Goffart <ogoffart@woboq.com>

Clarified documentation for QStandardItem::parent()

Task-number: QTBUG-18785

Change-Id: I91ee8c3dc10c1484376a17e3f13794038d4a648f
Reviewed-by: Stephen Kelly <stephen.kelly@kdab.com>
Reviewed-by: Girish Ramakrishnan <girish.1.ramakrishnan@nokia.com>

Add QAbstractDeclarativeData::isSignalConnected hook

Similar to QAbstractDeclarative::receivers. This hook will allow
QObject::isSignalConnected(QMetaMethod) to return true when there are
QML-managed connections.

It's important that the hook is called from
QObjectPrivate::isSignalConnected(uint), since QML calls that
function.

Change-Id: I9c5e42f2b3e7f985af02905985a3a47101cdee05
Reviewed-by: Olivier Goffart <ogoffart@woboq.com>
Reviewed-by: Michael Brasser <michael.brasser@nokia.com>

tst_qtextscriptengine: make it compilable under MSVC2010

Source files which contain UTF-8 literals can not be compiled
by MSVC with Chinese/Japanese locale.

Change-Id: I5daa2e45c5e1ceb86da91e72288c24018c49c0f6
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
Reviewed-by: Lars Knoll <lars.knoll@nokia.com>
Reviewed-by: Konstantin Ritt <ritt.ks@gmail.com>

Fix compilation for XCB-without-Xlib

Commit 6481218da13539529b87626a352a17f42c57be61 broke the build by
removing the #ifdef XCB_USE_XLIB guards, but forgetting to port
Button1 to the XCB equivalent.

Change-Id: I9102feceb4deb0135ce0df318e46b35d1134a0b9
Reviewed-by: Girish Ramakrishnan <girish.1.ramakrishnan@nokia.com>

mkspecs: Use the right variable name to find the qdevice.pri

Commit 8b822825c5066957622194acf0fc267a6bf473fd introduced the
/get version but used the wrong variable name. Fix it by using
QT_HOST_DATA.

Change-Id: Ia4759b8c6ff2de9726f3aebae2f2f39c6644d4ec
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@nokia.com>

Don't try build devicediscovery if evdev is not available.

Fixes build on Android.

Change-Id: I1e6af972e8c29dd4c38773985e8f201064c2164f
Reviewed-by: Girish Ramakrishnan <girish.1.ramakrishnan@nokia.com>

Fix QIODevice warning when running rcc.

When opening a QFile on stdout, for example,
we must not call seek as it is a sequential device.
This has been flagged as a warning since commit Ie3a96d3a
and has resulted in spurious warnings being emitted.

In the case of opening a QFile in Append mode, QIODevice::open
already sets the position marker, so calling seek is redundant.
This is also true for the file engine's open function (called
through openExternalFile()), which also ensures the handle or
descriptor is repositioned appropriately.

Task-number: QTBUG-26104
Change-Id: I71040c399efe54e7538f54433368b432e959e08d
Reviewed-by: Kent Hansen <kent.hansen@nokia.com>
Reviewed-by: Joerg Bornemann <joerg.bornemann@nokia.com>

clean up specdir()

the only callers which used non-default arguments are gone now, so remove
the arguments entirely. this also enables us to re-enable result caching.

Change-Id: I62f76e17e531a4eeafddb9b29716ca0a0eb3dbea
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
Reviewed-by: Joerg Bornemann <joerg.bornemann@nokia.com>

fix -spec/-xspec argument re-assembly

the specs in Option are now only fixifiable absolute paths if an actual
path was passed on the command line - otherwise these are just names,
and we should pass them on verbatim.

Change-Id: I5b3535dadf2d6b2b3116331e6636f6e363a5361f
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
Reviewed-by: Joerg Bornemann <joerg.bornemann@nokia.com>

Add missing subdirs (the new QUrl unit tests were not compiled and run)

Change-Id: I1b39d92b8a14d5aeca957180858e1980a534894b
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
Reviewed-by: Giuseppe D'Angelo <giuseppe.dangelo@kdab.com>

Fix PCRE build under non-__GNUC__ compilers

PCRE's JIT has several paths that end in a #error under compilers that
don't #define __GNUC__.

This is because either
- those platforms were unavailable to PCRE devs so they were not tested;
- the #ifdef guards inline assembly fragments in GCC (AT&T) syntax;
- the #ifdef guards functions present f.i. in ARM's EABI and unavailable
  f.i. under WinCE.

This commit disables PCRE's JIT under ARM and MIPS unless __GCC__ is
defined. The MIPS #define from MSVC (_M_MRX000) is also dropped.

Change-Id: I59f959c321413845ffbdf1ac32740b400422e0ee
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
Reviewed-by: Björn Breitmeyer <bjoern.breitmeyer@kdab.com>
Reviewed-by: Giuseppe D'Angelo <giuseppe.dangelo@kdab.com>
Reviewed-by: Andreas Holzammer <andreas.holzammer@kdab.com>

Remove toRect_normalized().

Seems to have been unused since qt4's fde7475bcf9c10522a8170e6eb8fb9a8fadc21cd.

Change-Id: I7ec88fb4ca8c1ffda469a1c048b4fea17ff042bc
Reviewed-by: Samuel Rødal <samuel.rodal@nokia.com>

Fix warning when ibus is not available.

This is perfectly ok on a linux desktop, so no reason to abort when
setting QT_FATAL_WARNINGS.

Change-Id: I64ba987679195741631fb204af57ae3d283bb516
Reviewed-by: Friedemann Kleint <Friedemann.Kleint@nokia.com>

QPlatformWindow: Add Window masks.

Add functionality for window masks to QPlatformWindow, which is
required to implement QWidget::setMask() in order to fix
the regression in functionality from Qt 4.8.

Change-Id: I2c2d5629f0b4c6d90e52595ad70b13559aab1f41
Reviewed-by: Samuel Rødal <samuel.rodal@nokia.com>

Make QRegion not need to be friends with QVector

This hack was introduced in Qt 4.5 and the intention was to resize
(hopefully, shrink) the QVector without causing it to reallocate
memory. We can accomplish the same by reserving the size: until we
clear() or squeeze(), the container will not free memory.

Change-Id: I03a09537a617be0ce865eefa043005e4051e8706
Reviewed-by: Lars Knoll <lars.knoll@nokia.com>

make using testlib automatically "promote" all modules to -private

autotests often need private headers (especially with qpa headers now
being private) and have no compatibility requirements, so it makes sense
to just use the privates of requested modules.

this also suppresses the useless warning about using privates, in case
they are still explicitly specified.

Change-Id: I9e499bedcf6ef25777283ff1432cef7254e9093a
Reviewed-by: Girish Ramakrishnan <girish.1.ramakrishnan@nokia.com>

ignore mkspecs/modules-inst/

only generated files in there

Change-Id: I6eb327fc686af08d46d343066420525bc2e0dde5
Reviewed-by: Rohan McGovern <rohan.mcgovern@nokia.com>

Make QGtkStyle build with QT_NO_FILEDIALOG

Fix compilation failure with qconfig large

Change-Id: I8ae14f01879b94430dcbb5c85c61d14e922f6eb9
Reviewed-by: Lincoln Ramsay <lincoln.ramsay@nokia.com>
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@nokia.com>

Factorize the cmake test infrastructure to be reusable.

This makes it easy to add cmake module tests for all modules.

Change-Id: I303bf7674ca6ae7a8544488f96e8e02afbaa6ff0
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@nokia.com>
Reviewed-by: Rohan McGovern <rohan.mcgovern@nokia.com>

Avoid macro re-definition after 1838a6c2fb5e3b4368c871aa330d028b255b83af

Change-Id: I2631ff687881278ba42edacd237a2130c103b69f
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@nokia.com>
Reviewed-by: Rafael Roquetto <rafael.roquetto@kdab.com>

QIODevice: free memory when buffer is cleared

The QIODevicePrivateLinearBuffer does not deallocate any data on
readAll or clear. This fix will change the buffer so that
data is deallocated on clear, readAll and when read emptied the
buffer.

This is needed for QAbstractSockets that don't have
readBufferMaxSize set, as the buffer will grow but never
decrease in size when you read from it.

Change-Id: Iab42e40182f9ebe0739c99b2d1e820ce287dc931
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@nokia.com>

QAbstractSocket: Enable readNotifier on read from buffer

This is needed for the QSslSocket. When we read on that socket we will
only read from the QIODevice buffer to get the unencrypted data.
So when the readNotifier has been turned off on the plainsocket there
is nothing to trigger it to be turned on again.

This will add a readData with zero size when we have read everything
from the buffer. This is so that we get a call into the socket to
check if the readNotifier should be turned on again.

Change-Id: I3b63e33de007db823e964480903186eb1b8caac2
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>

QtNetwork: remove double buffering on sockets

Removes the readBuffer from the QAbstractSocket since data is already
buffered in the QIODevice.

Change-Id: I4e50b791fd2852455e526fa2c07089d4d3f0b2a4
Reviewed-by: Prasanth Ullattil <prasanth.ullattil@nokia.com>
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>

[QNX] Enable MT screen event handling on Blackberry

Consuming incoming screen events through the BPS event
dispatcher on its current incarnation has proven to be
very inefficient since it cannot put up with the screen
event throughput. This patch enables the screen event
thread also for Q_OS_BLACKBERRY platforms, in addition
to stock QNX. This behavior can still be configured
through the project file, though.

Task-number: QTBUG-26177
Change-Id: I98a3b22549c9fbf0c16b5a8c39a55c1eaa6ec1f4
Reviewed-by: Sean Harmer <sean.harmer@kdab.com>
Reviewed-by: Kevin Krammer <kevin.krammer@kdab.com>
Reviewed-by: Giuseppe D'Angelo <giuseppe.dangelo@kdab.com>

Fixed inputMethodQuery() for Qt::ImHints

to return QWidget::inputMethodHints()

Change-Id: I46735c553e0cb4689cd0c53a69d07ed61ba56bf6
Reviewed-by: Joona Petrell <joona.t.petrell@nokia.com>

QTBUG-26035: Remove positive sign from start of string

Updated removeGroupSeparators(QLocalePrivate::CharBuff *num) so that it
removes also positive sign ('+') at the start of the string. Auto test
included.

Task-number: QTBUG-26035

Change-Id: I8e0e071d6c682d9192a8c6bb2f282510e21b3c48
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
Reviewed-by: Denis Dzyubenko <denis.dzyubenko@nokia.com>

fix somewhat common edge case in $$shadowed()

if source and build dir are direct children of the common root and we
are shadowing the top-level source dir, there is of course no trailing
slash to match.

Change-Id: I8a34a6a72d16cb21d77d056e037235af9b32a008
Reviewed-by: Davide Pesavento <davidepesa@gmail.com>
Reviewed-by: Andreas Hartmetz <ahartmetz@gmail.com>
Reviewed-by: Joerg Bornemann <joerg.bornemann@nokia.com>
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@nokia.com>

Use valSeen since it results in unused warning otherwise

Change-Id: Iccfa4b895d9e50227efa1747ea20ce07ce70327e
Reviewed-by: Rohan McGovern <rohan.mcgovern@nokia.com>

Don't re-evaluate the lib location.

Rely on the DESTDIR variable being set correctly by qt_module_config.

Change-Id: I1a166124024722ec5a189a7402b38646179aa890
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@nokia.com>

Fix the DEFINITIONS for Qt modules.

As the DEFINITIONS to be used for QtAddOns is different to essential
modules, rely on the logic in qt_module_config setting this variable
correctly.

Change-Id: I64485ccd6df093216cac4a97fb1cfaac0122a218
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@nokia.com>

QGraphicsItem::setAcceptsHoverEvents is deprecated, replace it

It is replaced by setAcceptHoverEvents (note the lack of s), even
though the original name made more sense -- "set (if this item)
accepts hover events".

Change-Id: Ia6137c0e37b0a5932836f1d08fb8fa436f736eed
Reviewed-by: Lars Knoll <lars.knoll@nokia.com>

Replace the QGraphicsItem deprecated transforming functions

The scale(), rotate() and translate() functions are replaced with
QGraphicsItem::setTransform.

Change-Id: Icb81c71b1513c049e2fd607995ca3a868108ee30
Reviewed-by: Lars Knoll <lars.knoll@nokia.com>
Reviewed-by: Andreas Aardal Hanssen <andrhans@cisco.com>

Small documentation fix about the text streams default encoding.

It's UTF-8, not Latin1 on most systems nowadays. Only Windows
still living in the past...

Change-Id: I70f1bd7a49bed6dcc8e39bbc0f0613475791afdb
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>

QSKIP test that is blocking unrelated changes in CI

The test looks vulnerable to misbehaviour if the working directory
contains unexpected files and folders. As it's already skipped on
Mac, skip on linux as well to unblock the CI.

Change-Id: Id2e48ea455eb77e36c4f9d899885e101f674c0a3
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>

We shouldn't have a Makefile here.

Change-Id: If55952b4aa9b95ff9311c1b7338b77af62a5c503
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>

Remove the font codecs

These codecs have only been used for XLFD based fonts.
These are not supported anymore by Qt 5.

Change-Id: I7dc083f2efcd42363b144b24bd62c169d83390cf
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>

The codecForTr() should be utf8

tr() assumes utf8 as input encoding, not
latin1.

Change-Id: If834f8c169bdb431d78713d14b03542d3a7ca8e4
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>

Fix docs mentioning Q_EXPORT_PLUGIN2

Fix all remaining places where Q_EXPORT_PLUGIN2
was being used in the documentation.

Change-Id: I7be67b83c18545d0e74f250b4b26583444b01909
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>

Use the new plugin system

Convert the last remaining three plugins over from
the old plugin system.

Change-Id: I355e6bb068ec4afb58a2ee9542f86e2913b3851d
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>

Fix return type of isActiveConfig("host_build")

The return type of isActiveConfig is a boolean, so return that when
host_build is queried.

Change-Id: I6d1420b49b09e51442c4b2d482e2f19b165081d8
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@nokia.com>

Make qtbase compile with QT_NO_DRAGANDDROP

Change-Id: Ief16e435af6e6d246b84505a1c1208994c7b0b38
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@nokia.com>
Reviewed-by: Jeremy Katz <jeremy.katz@nokia.com>