... instead of scoping the defines in qconfig.h, which relied on the
Q_PROCESSOR_xxx defines and meant that we had to include qconfig.h
after qprocessordetection.h, which added a whole bunch of other
dependency issues.
We now let configure write QT_COMPILER_SUPPORTS_xxx to qconfig.h as
before, without any scoping, and then undefine the ones that don't
apply for the given processor. This means we need to include
qprocessordetection.h before qcompilerdetection.h in qglobal.h,
but the former does not depend on the latter, so this should be
fine.
Change-Id: If00c00d405463e9626fa0f7f5e6b17f68778904f
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@digia.com>
The configure script does not actually produce a "confclean" make target.
Task-number: QTBUG-27735
Change-Id: Ia0f9e33e50c35cc6bb2941853e518a40fc9edaee
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@digia.com>
The warning is more likely to be seen if it's closer to the end of the
output.
Also report whether we found xkbcommon in the main output. Note that
xkbcommon is used by Wayland too, so it's not dependent on QPA or on
the XCB backend.
Change-Id: I143327eea4e17fa06bc7c24c677ae0bd00e65711
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@digia.com>
Keep options sorted and use indentation to group together related
options (QPA backends, image formats and SQL drivers).
Change-Id: I97d330a13a4daa4567ff741dc26e11f458ae7f53
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@digia.com>
Neither /dev/null nor NUL work with Qt on Windows. For now, use
an empty file for the cases where qmake is used with /dev/null.
I filed a bug for this:
https://bugreports.qt-project.org/browse/QTBUG-30562
Change-Id: If5351214ae5a0ebe50ae46b155c327ca0dc59f98
Reviewed-by: Alvaro Burnett <alvaroburnett@gmail.com>
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@digia.com>
A gap after option name is needed to select it with double-click.
Also I added padding periods for -testcocoon because it has no them before.
Task-number: QTBUG-30589
Change-Id: Ib5b970f9b17cad43609fbc53dd05a995aaf29b45
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@digia.com>
For now there's only one architecture per host/target, but
this will change once we start detecting the CPU features
for both device and simulator on iOS.
For convenience we set QT_CPU_FEATURES to the resolved
value of the current architecture, so that simd.prf still
can use QT_CPU_FEATURES directly.
Change-Id: I28e8b339a5c30a630e276165254dba09a3da6940
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@digia.com>
So that the user doesn't have to pass -no-pkg-config and -nomake, when
we already know that those flags are needed and that the build will break
without them.
Change-Id: Ic07e02bc1800f177cf09f704104c1a76bfc50aa2
Reviewed-by: Richard Moe Gustavsen <richard.gustavsen@digia.com>
Reviewed-by: Morten Johan Sørvig <morten.sorvig@digia.com>
QT_CPU_FEATURES is internal and only used by simd.prf, which is
also internal. This matches the available CPU features which are
detected using config tests and written to CONFIG in qmodule.pri.
Change-Id: I8eb35448e2954a54c228d3617f29afc0283a7db5
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@digia.com>
We don't need it. Let linux-g++ be the default on all Linux builds,
period.
Task-number: QTBUG-30590
Change-Id: I26c73bf4f054684763b64ef5651b3488363ea7a1
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@digia.com>
The default lib and include dirs are resolved using the target
compiler and with a sysroot, so they are not relevant for host
builds.
Change-Id: Iceb2eb865d0732b9a6f5896ad126200ae8e8a04e
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@digia.com>
We now have host_build to distinguish the two, and we load
qconfig.pri from both the host and the target mkspec, with
host_build set correctly.
Change-Id: I8b8b80d5487d10bb1d4585d27d10300f609a7775
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@digia.com>
We depend on Xcode for building Qt itself and user application on Mac OS.
The user may have an Xcode install that is not set up properly, in which
case we would fail compilation in mysterious ways. Instead we try to
detect misconfigured or missing Xcode installs as early as possible.
We try to detect if an Xcode install has not been chosen yet, and
if the user has not accepted the Xcode license agreement. We need to
do these checks both in configure, as early as possible, and in mkspecs
on Mac OS, as we need to error out if the user tries to build an app
with the Qt SDK, but with a broken Xcode install.
Change-Id: I4e3a11077a61dc5d4ee2c686d01044a9bb2c1c79
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@digia.com>
We always use the xcodebuild/xcrun/xcode-select binaries in /usr/bin,
as these will dispatch to the right binary based on what Xcode version
has been chosen using xcode-select -switch. This fixes an issue where
a tool was in the path from another Xcode installation. We can rely on
the tools as they are present on a clean Mac OS install.
Change-Id: I1d3cc1e92604f9be6d6f14639cb6322234edd696
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@digia.com>
This means we have to bump the deployment target to Lion (10.7), as the
LLVM 'libc++' C++ standard library does not support Snow Leopard (10.6).
For iOS the deployment target has to be bumped from 4.3 to 5.0, but we
don't enable C++11 by default yet as it's not tested enough on iOS.
Users who wish to deploy to 10.6 need to build their own Qt,
passing -no-c++11 to configure.
Change-Id: I7b5d20ab002db889d1091a4b7ff600f62caa7f06
Reviewed-by: Morten Johan Sørvig <morten.sorvig@digia.com>
We used to disable NEON completely, as the iOS toolchain does not
handle the GAS syntax of the pixman draw-helpers. But we can limit
the disabling to just the draw-helpers, which means we get NEON
optimization of eg. QImage and QString.
Change-Id: If350b06ce521cca8b24468be5a168ff21e9e7124
Reviewed-by: Richard Moe Gustavsen <richard.gustavsen@digia.com>
We're trying to parse GCC output, so let's make sure that they are in
English. I've seen some reports that "search starts here" was
translated to some locales.
Change-Id: If09b1f45607f65d054496db65418e413b8aa8d48
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@digia.com>
the de-duplication has the side effect of sorting, which is a very bad
idea: x-compilers tend to append the host library paths at the end, and
we really want them to stay at the end.
and the lists should have no duplicates to start with. should we find a
compiler which breaks this assumption, we can use qmake's $$unique()
strategically.
Change-Id: I01560e3c33736c2dfffdb05d5c960c492439c946
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
evdev support in eglfs is now guarded by QT_NO_EVDEV, so this
check is no longer required. This allows eglfs + tslib to be
a supported combination.
This reverts commit a95e396a83.
Change-Id: Icf7c15121b7eca1131d412b05b956cd5d7f189ae
Reviewed-by: Sean Harmer <sean.harmer@kdab.com>
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@digia.com>
the dlls being in lib/ is kind of an accident (a side effect of how
qmake builds them). in fact, the libdir should be entirely irrelevant
for windows deployments (and indeed, the SDK is delivered with dlls only
in bin/).
Change-Id: If47e72b24774721a61ba63847f6132f88ff110be
Reviewed-by: Friedemann Kleint <Friedemann.Kleint@digia.com>
Reviewed-by: Jocelyn Turcotte <jocelyn.turcotte@digia.com>
Otherwise, we end up with an empty QT_LFLAGS_SQLITE and the plugin
won't link.
Change-Id: I026f60bf9cd075218dbe2888fbb7fc82782b27ca
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@digia.com>
I don't need that warning on my Mac.
Change-Id: I18c06135ba88a037103fdda0982976f4a87c553e
Reviewed-by: Samuel Rødal <samuel.rodal@digia.com>
Reviewed-by: Gatis Paeglis <gatis.paeglis@digia.com>
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@digia.com>
Since c45595d648, we use the toolchain
based on the XCode SDK, not the one in $PATH. Turns out that the Clang
that comes bundled with XCode 4.6 reports its version as
Apple LLVM version 4.2 (clang-425.0.27) (based on LLVM 3.2svn)
Instead of "Apple clang". So we need to match for (clang|LLVM).
Extended regular expressions (with -E) were necessary because the sed
that comes with Mac OS X is apparently broken and will not work with
\(clang\|LLVM\). GNU sed accepts -E as an alias for -r, meaning
extended regexps.
Change-Id: I5a15de30721216b086c3d39a080cc6496c503985
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@digia.com>
Reviewed-by: Tor Arne Vestbø <tor.arne.vestbo@digia.com>
There is no logic in configure to detect compiler capabilities
in the host- and cross-compilers separately, so if the
cross-compiler has more capabilities than the host compiler, the
compilation will break. This was the case for c++11 on Mac, which
is supported by the Android cross-compiler but not by default on
the host. There is a fix planned to enable c++11 on Mac, so
this is a temporary patch to work around the problem by disabling
c++11 explicitly until it has been fixed.
Change-Id: I2048dc7f63991c97b11b3980ac91292d2c9b7ce4
Reviewed-by: Morten Johan Sørvig <morten.sorvig@digia.com>
this is cleaner than having it parse qmake project files.
the only remaining built-in version extraction is the fallback to
qglobal.h needed for bootstrapping.
as a "side effect", this fixes the build of modules with mismatched
versions centralized in .qmake.conf, as this was simply not handled so
far.
the -mkspecsdir syncqt option goes away, as there is no use case for it
any more.
Task-number: QTBUG-29838
Change-Id: I6912a38f0e93a26bc267a9e3d738506fd3ad431b
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
Reviewed-by: Jocelyn Turcotte <jocelyn.turcotte@digia.com>
Reviewed-by: Joerg Bornemann <joerg.bornemann@digia.com>
It is needed for implementing the shortcut functionality in the xcb platform
plugin and for the compose key input context plugin.
As announced on the wayland-devel mailing list - the libxkbcommon 0.2.0 is
the first grown-up release of the library (Tue Oct 23, 2012). [1]
[1] http://lists.freedesktop.org/archives/wayland-devel/2012-October/005976.html
Change-Id: Id5d45e1a5afe49cf9ec5312318bd173f5a067f62
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@digia.com>
Reviewed-by: Samuel Rødal <samuel.rodal@digia.com>
it's weird that the output from the two variants is differing, and that
after building qmake it appears to hang.
Change-Id: I2ac3ace11e958effe787b13e1300eb1d2839ae98
Reviewed-by: Joerg Bornemann <joerg.bornemann@digia.com>
7de9d3709 broke them, because suddenly every -l* switch was parsed as a
library. fix this by re-arranging how the options are parsed.
this obsoletes d7ab351cdd as well, by being more generic.
fwiw, this syntax is stupid to start with, because all unknown -l*
options are implicitly libraries and create confusing configure
failures. emulating the compiler/linker command lines isn't such a great
idea ...
Task-number: QTBUG-29174
Change-Id: I11bac7a6f458664dff8cbe57ed9cd33a08d5e9ec
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
Reviewed-by: Joerg Bornemann <joerg.bornemann@digia.com>
note that the value is written verbatim to the qmake file, so additional
quoting may be required on the command line.
Task-number: QTBUG-30102
Change-Id: I02ca9a44fae82b6932982e6385508b8a304cc1e7
Reviewed-by: Joerg Bornemann <joerg.bornemann@digia.com>
Since we use the Clang from Xcode's toolchain now, the OS version is
not relevant.
In practice this means we will use clang for Xcode 4.2 and up, which
means it's possible to use clang also on Mac OS 10.6 (Snow Leopard),
where Xcode 4.3 is not available.
Change-Id: I9817e237cdd82d10b93aaaa3c90e35767cdca751
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@digia.com>
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
For Mac OS X we currently specify build tools without an absolute path,
which means we end up using the ones in /usr/bin. This is wrong, we
should be using the tools from the toolchain of the chosen SDK.
For iOS we do specify an absolute path, by resolving the toolchain
path in the iOS makespecs.
To solve the situation on Mac OS X, we move the logic of resolving the
toolchain path to sdk.prf, and share it between OSX and iOS.
For configure we need to duplicate some of the logic from sdk.prf, as
configure pulls out QMAKE_CC and QMAKE_CXX for running some initial
tests and building qmake. The new macSDKify function also solves
the issue of missing sysroot and deployment version in the flags.
Change-Id: Ib1d239c9904cf3ccee5214b313cf6205869a1462
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@digia.com>
Is needed now that we write QMAKE_MAC_SDK with a !host_build scope
in qdevice.pri.
Change-Id: I298cc660b496460190337c175aef684a5522d5cb
Reviewed-by: Tor Arne Vestbø <tor.arne.vestbo@digia.com>
qdevice.pri is not target/host specific yet, so we were loading
it and overriding QMAKE_MAC_SDK to iphonesimulator eg., even
for host tools such as moc.
Change-Id: I10277e60e1da84dda239e32a6f19b40dc48f084a
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@digia.com>
Reviewed-by: Morten Johan Sørvig <morten.sorvig@digia.com>
Version 3.1, shipped with Xcode 4.3, has the same issues compiling Qt's
AVX code.
Change-Id: Icb778fbd9d61f01aa84365661af050c9442d4d7f
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
