Instead of using a define to rename the user's main() function during
compilation, we leave the user code alone, and inject our wrapper one
step earlier in the process, at the application entry point 'start'.
This entry point is provided by crt1.o, which is normally linked into
the application automatically. The start() function sets up some state
and then calls main(), but we change the start() function to instead
call our main wrapper.
Instead of shipping our own crt1 binary/sources, we make a copy of
the appropriate crt1.o at build time, and modify its symbol table in
place. This is unproblematic as long as we keep the same length for
the wrapper function name, as the symbol names are just entries in
the global string table of the object file.
The result is that for the regular Qt use-case the user won't see
any changes to their main function, and we have more control over
the startup sequence. For the hybrid use-case, we no longer rely
on the fragile solution of having our back-up 'main' symbol in
a single translation unit, which would break eg with --load_all,
and we don't need to provide a dummy 'qt_user_main' symbol.
OSX 10.8 and iOS 6.0 introduced a new load command called LC_MAIN,
which places the state setup in the shared dyld, and then just
calls main() directly. Once we bump the minimum deployment target
to iOS 6.0 we can start using this loader instead of LC_UNIXTHREAD,
but for now we force the classic loader using the -no_new_main flag.
There's also a bug in the ld64 linker provided by the current Xcode
toolchains that results in the -e linker flag (to set the entry
point) having no effect, but hopefully this bug has been fixed
(or Apple has switched to the LLVM lld linker) by the time we
bump our deployment target.
Change-Id: Ie0ba869c13ddc5277dc95c539aebaeb60e949dc2
Reviewed-by: Tor Arne Vestbø <tor.arne.vestbo@digia.com>
The cmake directory may not be $PREFIX/lib/cmake, but
instead $PREFIX/lib/<arch>/cmake. Getting the PATH of such a
directory will not lead us to $PREFIX/, but to $PREFIX/lib.
Use a relative calculation instead.
Task-number: QTBUG-33223
Change-Id: Ice4e0f859ab1df238bad4eb942f073e84dd86cc3
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@digia.com>
Reviewed-by: Lisandro Damián Nicanor Pérez Meyer <perezmeyer@gmail.com>
Reviewed-by: Stephen Kelly <stephen.kelly@kdab.com>
Commit a7ba0ad93e introduced the -Wno-
language. The warning about deprecated functions, variables or types is
-Wdeprecated-declarations.
Change-Id: I6d269851afefc6a3fc3bf6599c3c702eb164245e
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@digia.com>
the redundancy is entirely unnecessary at this point.
the cmake problem that prompted the revert has been fixed long since by
commit 4f5f9331d9.
This reverts commit 215f137e29,
thus restoring 6d61dfdbb74a2055438b999c6962f89cc3388eea..
Change-Id: I94749dc18d924163e9c9add500078d88ccd00ffc
Reviewed-by: Stephen Kelly <stephen.kelly@kdab.com>
It sounds like a good thing to have this warning, but for
future-proofing we can't have it. The system libraries might change
and add deprecation marks (OS X does that often). If they do that, we
don't want poor developers to have to fix all warnings before they can
build Qt again.
Change-Id: I4ff317da0de596c470bb1efe6e59bcf70aeec8fc
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@digia.com>
This allows us to go back to older versions of Qt with newer compilers,
that didn't exist when those versions were released. It also allows
someone upgrading their compiler and not being faced with having to fix
all warnings before Qt compiles.
This commit whitelists the following compilers:
* Apple Clang versions 4.0, 4.1 and 4.2 (OS X only)
* Intel Compiler versions 13.0, 13.1 and 14.0 (Linux only)
* GCC versions 4.6, 4.7 and 4.8 (all OS)
Notably, Clang on other other OS besides OS X and MSVC are missing.
Change-Id: I665160d40a59336da1904f2a6c1eda543e592b48
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@digia.com>
Due to forcing java 6 we need to pass the
bootstrap jar file (android.jar).
https://blogs.oracle.com/darcy/entry/bootclasspath_older_source
Change-Id: I530a7e2a7df40813011a6dde93d6ccc3aaaa61d6
warning: [options] bootstrap class path not set in conjunction with -source 1.6
Reviewed-by: Eskil Abrahamsen Blomfeldt <eskil.abrahamsen-blomfeldt@digia.com>
This is new in CMake 2.8.12 and replaces the old properties
matching IMPORTED_INTERFACE_LINK_LIBRARIES_<CONFIG>.
Change-Id: I5d4c454972f2535f6792e95718c73d80c56ac24c
Reviewed-by: Stephen Kelly <stephen.kelly@kdab.com>
This makes the diff clearer when adding plugin information
in a followup commit.
Change-Id: I857d9f71b08074f2ffa2f852ad72e5dd975adc3e
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@digia.com>
Reviewed-by: Stephen Kelly <stephen.kelly@kdab.com>
The first word of the .name variable of each extra compiler is
used as a key in a container that keeps track of them. See also:
qmake/generators/win32/msvc_objectmodel.cpp: VCProjectSingleConfig::filterForExtraCompiler
qmake/generators/win32/msvc_vcproj.cpp: VcprojGenerators::initExtraCompilerOutputs
Task-number: QTBUG-32912
Change-Id: I7ea5c58884db559621f50740075b7f2e4e3ef7f8
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@digia.com>
Reviewed-by: Joerg Bornemann <joerg.bornemann@digia.com>
under normal circumstances, any errors will be noticed already by the
pkg-config --exists call, which is silent anyway. therefore this doesn't
change anything in normal qmake usage.
however, lupdate's and creator's evaluators skip the --exists calls and
subsequently invoke the normal query, which then prints useless noise to
the terminal.
Task-number: QTBUG-28159
Change-Id: I536412060f3830aafeb0587f855cd6af11227bca
Reviewed-by: Joerg Bornemann <joerg.bornemann@digia.com>
qmake expects the generator to be the same for each node in the tree
of subdirs, including the leaf projects, which caused failures when
qmake tried to recurse out to the leaf projects and run 'make', when
the leaf project was an Xcode project.
We now wrap the Xcode project in a meta-makefile that just
calls out to xcodebuild to do the actual work. This allows us
to get rid of the hacky generator detection, and use the macx-xcode
mkspec instead of setting the generator, which is much cleaner.
Change-Id: I2fed6a4dd6343b6a320eb459ecae824553bff459
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@digia.com>
Reviewed-by: Richard Moe Gustavsen <richard.gustavsen@digia.com>
sed versions other than the GNU one often default to being POSIX-compliant,
in which case "+" (with and without escaping) is always an ordinary
character.
Achieve the same functionality in a way that works with both GNU and BSD
seds by using "xx*" insted of "x\+".
Change-Id: I1d2576a8c0e17b31f01a44d9632c57991e53780d
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@digia.com>
To instrument a Qt application or library with the gcov coverage
tool, do `CONFIG+=gcov' in the application .pro file.
To instrument Qt itself with gcov, use the `-gcov' configure
option.
Change-Id: If24e91d95318609b0df1a76ed6d679bd92bcaab2
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@digia.com>
Reviewed-by: Frederik Gladhorn <frederik.gladhorn@digia.com>
Change-Id: I68d087b15839418008db5bf1c0c76ca303245519
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@digia.com>
Reviewed-by: Stephen Kelly <stephen.kelly@kdab.com>
The $$path may already be absolute, so prepending a slash may
result in //usr/lib, for example.
Task-number: QTBUG-32570
Change-Id: If7a4f6fbec0216404cfe48c1da62d21d75b3e272
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@digia.com>
Reviewed-by: Stephen Kelly <stephen.kelly@kdab.com>
... as that causes debug+release installs to overwrite each other's
postprocessed files.
introduces CONFIG+=sliced_bundle, which instructs qmake to create
file-by-file install commands. we don't know whether people are not
putting files outside qmake's knowledge into the bundle build dir, so
this mode is not necessarily backwards-compatible, and thus off by
default.
Task-number: QTBUG-28336
Change-Id: I23e90985ccd3311f0237ed61aadca6d7ed8325b7
Reviewed-by: Joerg Bornemann <joerg.bornemann@digia.com>
Some packagers don't want to install the private headers.
Check the existence of private headers only if the 'Private' component
is specified when finding the package.
Task-number: QTBUG-32466
Change-Id: I1fdbfb25e8ce485cd051564b937f766b2733741a
Reviewed-by: Stephen Kelly <stephen.kelly@kdab.com>
Use "_" instead of "-" in variables so variable replacement works
properly.
Change-Id: I2b17dca8f2351bc0933c165017f3fbb9393b0514
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@digia.com>
For developer builds, there is no need to run the test a second time.
Change-Id: I3564874cb2e9d6cc243e25a89ecd7f89df23b0bd
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@digia.com>
This gives us better consistency across the Qt ecosystem.
Change-Id: Ie12ebb6e8c826ed2e0445eb37de0b79595da41c2
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@digia.com>
The test should still be run, even though it is insignificant.
Change-Id: I6a3853e2b0e9670152b4f329dbceed2986a7e008
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@digia.com>
the forward-referenced directories don't exist yet, so we get pointless
warnings. in fact, this is why we do a multi-pass build in the first
place, and consequently using indexes during the first pass is
illogical.
Task-number: QTBUG-32152
Change-Id: I66bf6b43238827e87cb8bf6932d581b808c1032d
Reviewed-by: Martin Smith <martin.smith@digia.com>
the previous attempt broke ActiveQt, as it actually has public modules
without headers (they are provided by a common base module).
so explicitly mark the internal modules as such instead of applying
heuristics.
Change-Id: I8d8a2ee66f02c3444da2036a497e7f382f089f62
Reviewed-by: Joerg Bornemann <joerg.bornemann@digia.com>
Google moved dx.bat into a new build-tools/VERSION folder
meaning our dx.bat no longer found dx.jar. Fix this by
passing into our dx.bat, the location of the real dx.bat
Removed hardcoded 17.0.0 and %ANDROID_BUILD_TOOLS_REVISION%
path searches.
Change-Id: I91c12c01745d6f12edbd126102b8f06eba291402
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@digia.com>
Reviewed-by: Eskil Abrahamsen Blomfeldt <eskil.abrahamsen-blomfeldt@digia.com>
This reverts commit 1ef74a763a, as
assembler files in SOURCES break compiling with -pch, as we don't create
a respective PCH.
instead, compile assembler code with QMAKE_CC, not QMAKE_CXX.
the reason why this change is needed in the first place is not clear to
me, but i guess that CXX defines some c++-related macros when
preprocessing the file, which breaks further down the line. this is
counter-intuitive, as the g++ frontend should treat the same sources the
same way as the gcc frontend (differences should be limited the the ld
invocation).
Task-number: QTBUG-29765
Change-Id: Ic0116b3a5fa621f12ac41cadf3062ff00b538e85
Reviewed-by: Rafael Roquetto <rafael.roquetto@kdab.com>
Reviewed-by: Joerg Bornemann <joerg.bornemann@digia.com>
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
if a private module was used without the suffix, it would not add any
include paths, but the library would be still added. as long as the
includes were written as <Module/private/Header>, this would not become
visible, as the public modules would add the common include path ...
however, this soon won't be the case for mac frameworks any more. this
change makes the problem visible early on.
Change-Id: I8b1a20313ad736cb49507f07fa623e9aa812f651
Reviewed-by: Joerg Bornemann <joerg.bornemann@digia.com>
The Qt CI system runs the unit tests after installation, but
with the qmake in the build directory. This means that the
installed content is not unit tested. Add an additional cmake
unit test to test the files in the install location.
The new test is marked insignificant for now until the true
effect on the CI system is known.
Task-number: QTBUG-27315
Change-Id: If9f12e88cfc741946cfabc25dbf789a11a2af4b8
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@digia.com>
We haven't fully agreed on whether headers should be in $includedir or
not for builds with frameworks, since frameworks carry their headers
inside. However, this change came too late for Qt 5.1 for us to assess
the potential impacts -- it's known to break at least the cmake files we
ship.
So restore the build to the way it was.
This is a partial revert of 6d61dfdbb7.
Partial because the old behavior was partially broken: it did not
install headers for release builds. Now we always install (and in
debug-and-release mode, we might do it twice).
Task-number: QTBUG-32134
Change-Id: Ib84879c5a148d3717d16a7a90b2f5735fb5d80be
Reviewed-by: Sean Harmer <sean.harmer@kdab.com>
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@digia.com>
Reviewed-by: Lars Knoll <lars.knoll@digia.com>
only the include statements found in public headers are constrained to
work with this flag. our own c++ files and private headers can use other
styles, which this flag breaks.
Change-Id: Icb1ced17dc438083731049788ac28349c87ba7ef
Reviewed-by: Joerg Bornemann <joerg.bornemann@digia.com>
Reviewed-by: Kai Koehne <kai.koehne@digia.com>
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@digia.com>
they are a somewhat different kind of private headers, but follow
generally the same logic.
Change-Id: Ic6f42ed7061dde2ffd0e32b1d713354b58a20970
Reviewed-by: Joerg Bornemann <joerg.bornemann@digia.com>
Reviewed-by: Kai Koehne <kai.koehne@digia.com>
the entry for the normal headers already ensures that there is the
correct version symlink. having an entry for the private headers as well
is pointless, and in fact clobbered the symlink for the actual library.
Change-Id: I2403761bf006b7bfa490ce85c7b0e46d5ef203c0
Reviewed-by: Stephen Kelly <stephen.kelly@kdab.com>
Reviewed-by: Tor Arne Vestbø <tor.arne.vestbo@digia.com>
the only case where we want to skip copying the bundled data (which is an
optimization only) is the debug sub-build when we are actually building
both debug and release.
Change-Id: I1f3f67ccd9a64033b133ffaf58639cd9f7107c27
Reviewed-by: Tor Arne Vestbø <tor.arne.vestbo@digia.com>
i see no particular reason why debug builds should still get
"normal" headers.
Change-Id: I3625648549e8c234a365bab26823190ed2219cdf
Reviewed-by: Tor Arne Vestbø <tor.arne.vestbo@digia.com>
there is no point in adding a structure we don't actually define.
Change-Id: Ica43123f17eca6ebd4b5b7ec2526ebabef31c82a
Reviewed-by: Tor Arne Vestbø <tor.arne.vestbo@digia.com>
it's cleaner, and it makes it possible to actually have a single else
branch.
Change-Id: I5ef917b678e2bd5a2face8ee19e942e5e952aa80
Reviewed-by: Tor Arne Vestbø <tor.arne.vestbo@digia.com>
don't mess with the -F linker flag manually. qt headers include other
headers via the canonical Module/Header syntax, which means that the
compiler also needs the -F flag. QMAKE_FRAMEWORKPATH does exactly that.
Task-number: QTBUG-29003
Change-Id: I5f4af1a462697cd6996c54436ccdb9fc2b216020
Reviewed-by: Andy Shaw <andy.shaw@digia.com>
Reviewed-by: Joerg Bornemann <joerg.bornemann@digia.com>
... except for MAKEFILE_GENERATOR = XCODE. This means the spec no longer
hard-codes g++, and will work regardless of whether the default spec was
clang or g++.
This require us to set QMAKE_XCODE_GCC_VERSION properly for GCC, so that
additional compilation flags passed by Xcode will match the actual
compiler used.
Task-number: QTBUG-31713
Change-Id: If65140a7471cd16f483036742f1d5b86d0485c52
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@digia.com>
Otherwise doing stuff like -spec macx-g++ when the default spec is clang
will not have an effect on the tools used.
Change-Id: Ia2769abfdd8c19f79d427b9f09707430e736305a
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@digia.com>
If Qt was built with C++11, it links to libc++, and we need all projects
that use Qt to link against the same library.
Technically, we could do QMAKE_LFLAGS += -stdlib=libc++, but that hasn't
been tested enough without also enabling C++11, so we keep the
relationship between the two for now.
Change-Id: Ic628bcbade60cc82f93707f372c2119c24d9dc8a
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@digia.com>
Reviewed-by: Liang Qi <liang.qi@digia.com>
The cmake variable for the mkspec dir must specify the source
location if used in the build dir, and must specify the install
location if used in the install dir.
Change-Id: I2fee8cd0c7198e9fc5cbb63972e20c75636672d1
Reviewed-by: Thorbjørn Lund Martsum <tmartsum@gmail.com>
Reviewed-by: Stephen Kelly <stephen.kelly@kdab.com>
the evaluator has the bug that function arguments are inherited.
work around that by passing an explicitly empty 3rd parameter to
qtAddTargetEnv().
proper fix upcoming in less critical branch.
Change-Id: Ic45cc890abaa6271985590d4ebe02c96bff6dec4
Reviewed-by: Kai Koehne <kai.koehne@digia.com>
Reviewed-by: Joerg Bornemann <joerg.bornemann@digia.com>
qtAddModule() always returns true anyway.
the real checking is done by qtAddModules() and qtAddLibrary() itself.
Change-Id: Ieed821acc36dc57ca52aec3e6b2dd6513be9b6c1
Reviewed-by: Joerg Bornemann <joerg.bornemann@digia.com>
unlike the .command, the .depend_command is not executed by make via its
chosen shell, but qmake itself via the system's native shell.
consequently, it needs different path separators and no make-escaping.
Task-number: QTBUG-31289
Change-Id: I480f815753632db6e8d4725f463f8a1fc59680a6
Reviewed-by: Joerg Bornemann <joerg.bornemann@digia.com>
nmake needs %-escaping in addition to $-escaping, not instead.
this has little practical impact, so it went unnoticed.
Change-Id: I144b6142eec0151d83a22e0ac5ead7b0415cdafa
Reviewed-by: Joerg Bornemann <joerg.bornemann@digia.com>
the vs ide executes the commands verbatim, so they must not be
make-escaped.
Task-number: QTBUG-31289
Change-Id: Ie73fd5c4da5527c2d10bc94ccdf60f8a1ca21351
Reviewed-by: Joerg Bornemann <joerg.bornemann@digia.com>
the precise syntax depends on what exactly the command is used for, so
we need to resolve it at the last moment. see followup commits.
This logically reverts commits 6f4ff81380
and 731e6bece5.
Change-Id: If285c91d7521069be86d32593b5c2ae2027b3038
Reviewed-by: Joerg Bornemann <joerg.bornemann@digia.com>
Change-Id: I840f963c3648d123b31f79aa2c8902c0ad74e982
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@digia.com>
Reviewed-by: Stephen Kelly <stephen.kelly@kdab.com>
instead, use the files directly from the source dir.
Change-Id: I03b728c66de6e03cade6dc153dcc78cea8e3f606
Reviewed-by: Jerome Pasion <jerome.pasion@digia.com>
Reviewed-by: Martin Smith <martin.smith@digia.com>
Reviewed-by: Joerg Bornemann <joerg.bornemann@digia.com>
We can combine the hybrid and non-hybrid use-cases into a single static
library if we are careful about which symbols are included in which
object files. By limiting the main() and qt_user_main() functions to
their own translation units, the linker will only pick them up if they
are missing at link time (the user's program do not provide them).
This technique is resilient to the -ObjC linker flag, which includes all
object files that implement an ObjectiveC class or category, but will
fail if the -all_load flag is passed to the linker, as we'll then have
duplicate symbols for either main() or qt_user_main(). The latter should
not happen unless the user provides the flag manually, and in the case
he or she does, there's ways to work around it by providing less global
flags such as -ObjC or -force_load.
Change-Id: Ie2f8e10a7265d007bf45cb1dd83f19cff0693551
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@digia.com>
Reviewed-by: Richard Moe Gustavsen <richard.gustavsen@digia.com>
Instead of force-loading the whole static library of the platform plugin
we tell the linker to look for the missing symbol qt_registerPlatformPlugin.
This symbol is provided by the same object file as the plugin's static
initializer, so the object file is included in the final binary and
the static initializer is run, resulting in the plugin registering with
Qt.
We could have marked the actual static initializer wrapper provided by
Q_IMPORT_PLUGIN(QIOSIntegrationPlugin) as undefined, but due to the C++
mangling this would look less intuitive on the linker command line than
the custom dummy function that we provide, which has C linkage.
Change-Id: I6805537e1f49260a41d48c555376964cb1fe75d8
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@digia.com>
Reviewed-by: Richard Moe Gustavsen <richard.gustavsen@digia.com>
the commands are already quoted appropriately for the shell.
Change-Id: I746bb5fba2cd6548c5dc7ef81087c69a200ecbb8
Reviewed-by: Joerg Bornemann <joerg.bornemann@digia.com>
in the actual specs, we also set 'gcc' for clang.
Change-Id: Ifc6b27d56596f34c944205795d665f545d090f80
Reviewed-by: Joerg Bornemann <joerg.bornemann@digia.com>
Reviewed-by: Jake Petroules <jake.petroules@petroules.com>
These directories are not currently part of the Qt installation for mac
frameworks.
Task-number: QTBUG-31641
Change-Id: Ifef372cc2ebb692f9ae5a7b1f8dba5f683d1e7eb
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@digia.com>
Reviewed-by: Stephen Kelly <stephen.kelly@kdab.com>
This is actually a list in CMake, not a value to be prepended
to paths. Specify the QT_SYSROOT instead to root the location
of include directories. CMake will soon get a CMAKE_SYSROOT
variable which will replace this.
Change-Id: I239f69f127f3676a3835aa4f29638f44ef209819
Reviewed-by: Volker Krause <volker.krause@kdab.com>
Reviewed-by: Stephen Kelly <stephen.kelly@kdab.com>
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@digia.com>
nonrelocatable adds the full uri to the exported type information
which is the correct thing to do for the qt plugins (and 99% of
the cases).
This way import bla.x 1.0 works correctly in the code model.
Change-Id: Ia06873dd8b2ea4627e3297a98e8df87275ceaf73
Reviewed-by: Alan Alpert <aalpert@blackberry.com>
Reviewed-by: Kai Koehne <kai.koehne@digia.com>
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@digia.com>
make the include dir in the source tree the "main" include path, as
that's where the majority of the headers is. then selectively add the
shadowed dirs.
Change-Id: I03ad13cfcf77175c141b94d41b1221740d851faf
Reviewed-by: Joerg Bornemann <joerg.bornemann@digia.com>
