Somehow qmake doesn't add the correct rules for the Android makefiles,
so the build fails when cross-compiling from Windows. The reason for
that is unknown (could be related to that "qt_android_deps" config, but
that isn't used anywhere in qmake or the buildsystem).
This isn't likely to be a problem, since there are no global installs of
Qt on Android.
Change-Id: I1d0f78915b5942aab07cffff140f95ce32324030
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
This way, it's possible to tell which applications and libraries depend
on the Qt private API and of which Qt library. Linux distributions can
use this information to decide which applications need to be recompiled
every time Qt itself is rebuilt.
This is done by scanning all class and struct definitions in the private
headers (we've already got the list from syncqt). I opted to add a new
script instead of modifying syncqt because then this can run in parallel
with the rest of the compilation, as opposed to during qmake
time. Another advantage is that it catches modifications to the headers
in between qmake executions.
Since this is already Unix specific, it should be no problem to use Perl.
This solution is limited to use of non-inline symbols of classes
declared in private headers. It will not catch free variables (such as
qsimd_p.h's qt_cpu_features), use of inlined functions or just plain use
of a class/struct for accessing its data members. However, this is
already better than nothing and should help Linux distributions quite a
lot. And there's no way to catch the latter issue anyway.
Change-Id: I049a653beeb5454c9539ffff13e3fff36400ebbd
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@theqtcompany.com>
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
The only reason I had used them in the first place was because C
preprocessor macros cannot call themselves recursively. But the magic
was too magic and caused issues with some builds, so let's choose the
safer option.
Anyway, this solution now works for all ELF architectures, independent
of the processor, whereas previously it was restricted to x86 and Linux/
FreeBSD. However, this does not apply to the assembly in
qversiontagging.h.
Change-Id: I42e7ef1a481840699a8dffff1404f032fc5cacb8
Reviewed-by: Lars Knoll <lars.knoll@theqtcompany.com>
the addition of qt's rpath belongs into qt.prf - even on mac.
so consolidate the two implementations.
as a nice "side effect", we get relative rpaths also on linux.
another "side effect" is that we don't unnecessarily add the qt rpath to
qt modules also on linux.
the qt rpath addition mechanism should not be responsible for setting
the policy who gets a relative rpath, so move the logic to higher-level
callers.
Change-Id: I52e8fe2e8279e7b1ac25fae758867a5cb1cafcf8
Reviewed-by: Joerg Bornemann <joerg.bornemann@theqtcompany.com>
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@theqtcompany.com>
This linker script is only enabled for systems with GCC or GCC-like
compilers, though technically it should work on the BSDs too (will
enable after testing). For regular modules, this declares one ELF
version "Qt_5" and places all QtCore symbols inside, then it declares
unused ELF versions "Qt_5.x" for each older minor release. For modules
declared "internal_module", all symbols are placed in version
Qt_5_PRIVATE_API.
The big advantage of an ELF version is that, when we do Qt 6, both
versions of QtCore could be loaded in memory without conflicts and all
symbols would be resolved to the correct library. No module can talk to
both at the same time, but this avoids mistakes of loading them
indirectly by plugins.
The extra Qt_5.x versions will be used in the next commit.
Change-Id: I049a653beeb5454c9539ffff13e3fe6f050fdf31
Reviewed-by: Lisandro Damián Nicanor Pérez Meyer <perezmeyer@gmail.com>
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@theqtcompany.com>
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
Defaulting to absolute_library_soname on configure -rpath is no longer
necessary as now we support @rpath install name ids on OS X and iOS.
This also sets QMAKE_SONAME_PREFIX to @rpath for Qt modules when built
with rpath configuration.
This makes Qt libraries relocatable on OS X. Qt SDK is not yet
relocatable though, because plugin locations (including cocoa plugin)
are still resolved using absolute path (see QTBUG-14150). Also, there
are several absolute paths hardcoded in qmake mkspecs pri files.
Task-number: QTBUG-31814
Change-Id: I36b9384cd69ac609608acbe2b3d5e0512317e0d6
Reviewed-by: Tor Arne Vestbø <tor.arne.vestbo@theqtcompany.com>
there is no need to make exceptions for install targets now, so instead
of abusing qt_no_install_library, introduce a new header_module flag.
Change-Id: I4ad7e301d1b60938b17e1dea732b1dbe3ff88a1a
Reviewed-by: Joerg Bornemann <joerg.bornemann@theqtcompany.com>
automatically set TEMPLATE=aux if qt_no_install_library is set.
Change-Id: Iccceda468da762b181fdd5c8e511bf6ed19af599
Reviewed-by: Joerg Bornemann <joerg.bornemann@theqtcompany.com>
just like in qt_plugin.prf, the DESTDIR setting is actually fixed per
module type.
Change-Id: I5837b5884699f0d50e4067733af8aacbab93bc42
Reviewed-by: Stephen Kelly <ske@ableton.com>
Test each include file directly, instead of doing a large #include. This
verifies that each header is compilable on its own. One big advantage of
doing it via a special compiler in qmake is that we skip pre-compiled
headers, which has hidden build errors in the past.
This solution is implemented by making syncqt produce a second list of
headers. This list is the same as the list of headers in the source
code to be installed, minus the headers that declare themselves to be
unclean, via the pragma:
#pragma qt_sync_skip_header_check
This mechanism is applied only for public libraries (skipping
QtPlatformSupport, an internal_module).
This test is enabled only for -developer-builds of Qt because it
increases the compilation time.
On QtTest: the library only links to QtCore, but it has two headers that
provide inline-only functionality by including QtGui and QtWidgets
headers (namely, qtest_gui.h and qtest_widget.h). If those two modules
aren't getting compiled due to -no-gui or -no-widgets to configure, we
need to remove the respective headers from the list of headers to be
checked. If they are being built, then we need to make QtTest's build
wait for the headers to be generated and that happens when qmake is
first run inside the src/gui and src/widgets directories.
Change-Id: I57d64bd697a92367c8464c073a42e4d142a9a15f
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@theqtcompany.com>
the function is used in our examples and code generated by qt-creator,
so the qt5-specific magic behavior is inappropriate. create a separate
function instead.
Task-number: QTBUG-44595
Change-Id: I4d72cc1e5cbfc274b3210520baa213f4c5479ca9
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
pkg-config .pc files use the raw target paths (and pkg-config patches up
-I and -L flags on the fly), so these files were actually already fine.
libtool .la files use the magic prefix = to denote the sysroot.
this works only with libtool 2.4+ (sept 2010).
qmake .prl files have no built-in sysrootification magic, but as they are
read by qmake, it's possible to put property references into them. this
makes them relocatable, both inside and outside sysroots.
Change-Id: I97236ac81e7aba4e4771d14a44cbf59144cc2d3e
Reviewed-by: Joerg Bornemann <joerg.bornemann@theqtcompany.com>
This sets the prefix for frameworks to "org.qt-project".
Applications keep using the default Xcode preferences
prefix.
Task-number: QTBUG-32896
Change-Id: I67384f643888f2de3dd8e36b9bce0f04ca4e16dd
Reviewed-by: Tor Arne Vestbø <tor.arne.vestbo@digia.com>
The change was made too late in the 5.4.0 release
cycle, and broke the Qt build and deployment in
several areas:
- macdeployqt
- OS X 10.7 builds
- shadow builds
This reverts commit c0a54efc40.
Change-Id: I1c1ad4901228f5516352ccdfa963e8ea2b5013b6
Reviewed-by: Morten Johan Sørvig <morten.sorvig@digia.com>
Reviewed-by: Gabriel de Dietrich <gabriel.dedietrich@theqtcompany.com>
Defaulting to absolute_library_soname on configure -rpath is no longer
necessary as now we support @rpath install name ids on OS X and iOS.
This also sets QMAKE_SONAME_PREFIX to @rpath for Qt modules when built with
rpath configuration.
This makes Qt libraries relocatable on OS X. Qt SDK is not yet relocatable
though, because plugin location (including cocoa plugin) is still resolved
using absolute path (see QTBUG-14150), also there are several absolute paths
hardcoded in qmake mkspecs pri files.
Task-number: QTBUG-31814
Change-Id: Ie9dffefcd2a946c1580293d433621c1adb7e06c4
Reviewed-by: Jake Petroules <jake.petroules@petroules.com>
with qmake now de-duplicating the paths properly, we should version
these QMAKE_BUNDLE_DATA entries as well, so we don't depend on the
symlinking of the regular headers being done first.
Change-Id: Idaa2ccc1ba9b5684b0c8d84f7f760735f54432e1
Reviewed-by: Morten Johan Sørvig <morten.sorvig@digia.com>
This follows the discussion at:
http://lists.qt-project.org/pipermail/development/2014-June/017225.html
Qt WebEngine will have a version of 1.0 when released with Qt 5.4.
The library name is currently libQt1WebEngine.so.1.0.0 but it should
rather be libQt5WebEngine.so.1.0.0 to represent Qt's major version
releases as a whole and not the major version of the module. This
prefix essentially expresses the module's dynamic linking
compatibility with other Qt modules.
This only makes sense if each major module release will be compatible
with a single Qt major version only.
All published modules currently already have 5 as their major version,
except qtenginio which doesn't use a Qt prefix, so this change has no
effect except for qtwebengine.
Task-number: QTBUG-30910
Change-Id: I894e7a367624c7fc263cf08104173a82eafd1439
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@digia.com>
Move it from bootstrap.pro into qt_module.prf so it will apply to any
other bootstrapped libraries, like libQmlDevTools.
Variable called "SPLIT_SECTIONS" because -fdata-sections could be added
in the future, if it proves to be a benefit.
Change-Id: I3fbb004f111620a84e58e9112e9bce3afd95631e
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@digia.com>
neither .prl nor .la files even contain include paths.
in .pc files, we assign the final path to start with, so there is
nothing to replace, either.
Change-Id: I919dfa01e0a1d0ef8ef1220174de1d33c66cedf9
Reviewed-by: Joerg Bornemann <joerg.bornemann@digia.com>
Added a define to suppress MSVC warning “C4996:
This function or variable may be unsafe.”
If the code is really "unsafe" then it is unsafe on
other platforms as well; so fixing these warnings just
for MSVC builds, would clutter the code and wouldn't help
in fixing issues that might exist on other platforms.
Using the same functions across all supported platforms
keeps the code clean and helps in writing code that is
safe across all platforms.
Change-Id: I470072eda4f8174bb911567ef3f061a3582ba449
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
Commit 773dd01 introduced a general mingw platform scope, which
is cleaner and more flexible than matching the spec name.
Change-Id: Ie3a9cb791a83f7c8a51bc4e23069190c452ab521
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
...unless the user passed the -no-sse2 option to the compiler.
[ChangeLog][Important Behavior Changes] Qt now automatically generates
code for processors supporting SSE2 on i386 platforms. To disable
this, pass the -no-sse2 option during Qt configuration. Since this
feature has been present on CPUs for 10 years and since Qt no longer
checks for runtime support for SSE2, we strongly encourage users to
leave the default setting on for best performance.
- For Linux distributions that must retain support for CPUs without
SSE2, we recommend doing two builds of Qt and installing the
SSE2-enabled libraries in the LIBDIR/sse2 directory. Tools,
plugins, and examples are not affected.
- See discussion on the Qt development mailing list:
http://lists.qt-project.org/pipermail/development/2013-November/014085.html
Change-Id: I7f9b1f58a9f66b6e5fe295bac15f87d34343695e
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@digia.com>
Reviewed-by: Olivier Goffart <ogoffart@woboq.com>
Reviewed-by: Lars Knoll <lars.knoll@digia.com>
it's bogus in the first place that the meta files contain windows paths,
but straightening that out is a prohibitive effort. so instead generate
additional s/// commands which take care of these paths.
fwiw, the generated s///i command is a gnu extension. but as we are
doing this on windows only where we are using our built-in sed command
anyway, this should be fine.
Task-number: QTBUG-33794
Change-Id: I46fcc598db12816ee56b5371ab184f6277eb3a22
Reviewed-by: Joerg Bornemann <joerg.bornemann@digia.com>
!build_all is a sufficient condition for bundling headers in both debug
and release configurations, as it means that we will likely make only
one of the builds, and that needs to be self-contained.
Task-number: QTBUG-34575
Change-Id: I1a19e4b619eeff207cfbfd50d3b761053aeaa667
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
instead of being magic attributes of the main modules, the privates
are now proper modules of their own.
this cleans up some code paths, is more mappable to other build tools,
and enables private modules to depend on other private modules.
note that the library path is needed even in the "empty" private
modules, as in the framework case that's where headers are found.
consequently, the modules need to be explicitly marked with the new
"no_link" flag. this required some reorganization of qtAddModule().
Change-Id: I8e4f44a609f8d639cc01bcb658256870a627eb63
Reviewed-by: Joerg Bornemann <joerg.bornemann@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>
... 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>
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>
... and introduce -hostlibdir configure option for symmetry.
the libraries built for the host have no business in the target prefix.
in principle this code would even support dynamically linked host
libraries, but that's currently unused.
Task-number: QTBUG-30591
Change-Id: I8e600fa4911a020fb0e87fbf7ef2f35647c7c4d5
Reviewed-by: Joerg Bornemann <joerg.bornemann@digia.com>
Reviewed-by: Ivan Romanov <drizt@land.ru>
... like qt_tool does. otherwise we get linker errors with debug builds
on windows.
Change-Id: I583f277ff3fb75c9fe5f305a6f1b5d066b840c07
Reviewed-by: Debao Zhang <hello@debao.me>
Reviewed-by: Joerg Bornemann <joerg.bornemann@digia.com>
Reviewed-by: Jonathan Liu <net147@gmail.com>
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@digia.com>
