this is a monster commit which does the following things:
- import the evaluator as-is from qt creator into qmake/library/
- integrate it into qmake's makefiles
- overwrite proitems.h with actual special types
- remove the parts of Option which are redundant with QMakeGlobals
- make QMakeProperty a singleton owned by Option::globals. the dynamic
handling so far made no sense.
- make QMakeProject a subclass of QMakeEvaluator, with relatively few
extensions
the changes to existing qmake code outside project.* and option.* are
minor. implementing the changes gradually would mean changing a lot of
code which will be just replaced in the next commit, so i'm not wasting
my time on it.
Change-Id: I9746650423b8c5b3fbd8c3979a73228982a46195
Reviewed-by: Qt Doc Bot <qt_docbot@qt-project.org>
Reviewed-by: Joerg Bornemann <joerg.bornemann@nokia.com>
this is preparation for adapting to a new evaluator.
Change-Id: I6fc59f5525735754a00afa6629fbfe257e84db97
Reviewed-by: Daniel Teske <daniel.teske@nokia.com>
Reviewed-by: Joerg Bornemann <joerg.bornemann@nokia.com>
this may have worked a decade ago, but now it only produces funny
Makefiles (and needs hacking main.cpp). the feature doesn't seem *too*
important, so just clean it out.
Change-Id: I50a60b0e30341f0b523e4a5731c770c9c1013f8b
Reviewed-by: Daniel Teske <daniel.teske@nokia.com>
Reviewed-by: Joerg Bornemann <joerg.bornemann@nokia.com>
don't read the spec from scratch for every library just to get
QMAKE_LFLAGS_RPATH. we can perfectly use our current project for that
purpose.
Change-Id: I4e408b3fd5de81652181df032aa53cd8f2f8f806
Reviewed-by: Joerg Bornemann <joerg.bornemann@nokia.com>
By putting object_with_source into CONFIG one could force qmake to
output each object file into the same directory as its source file
came from. This was a rather nasty work-around from Qt 3 times to
support source files with the same file name in a project.
Unfortunately this doesn't play nicely with shadow builds.
Change-Id: Ie79e14d36ba6eac4219edc14ea75ab6a96f9ea96
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@nokia.com>
instead of re-assembling a list from the variables, take the original
command line minus some explicitly stripped out options. this is way
less code and poses no synchronization problem between the two parts.
as a "side effect", variables obtained from $QMAKEFLAGS won't multiply
with each makefile nesting level, as the generated command line won't
replicate data obtained from the environment.
Change-Id: I5d1ce0f11efb338f60405529f9818910103b1b0e
Reviewed-by: Joerg Bornemann <joerg.bornemann@nokia.com>
this adds a .base "member" to the install "structure". if specified, only
this much is stripped from the front of each element of .files, rather
than the entire path, to obtain the target filename.
Change-Id: Ic39fcf71c4ad874ffabbbad113be9cdc6e3f7260
Reviewed-by: Joerg Bornemann <joerg.bornemann@nokia.com>
don't make a single string, but a string list which is join()ed in the end.
this is a tad slower, but the code is way easier to work with.
Change-Id: I1ff7168c2770998761a6081be8080c743ddc94a1
Reviewed-by: Joerg Bornemann <joerg.bornemann@nokia.com>
make a proper stringlist of commands, and join it in reverse order
only at the end. the reversal ensures that we can cleanly fold up
directory hierarchies we may build.
Change-Id: I9a241361588a6965283aec5258e1d622b35514e0
Reviewed-by: Joerg Bornemann <joerg.bornemann@nokia.com>
otherwise the second installation on unix would be bogus.
Change-Id: I162533ee262c6820e7e2d4710b5342cafecd9d59
Reviewed-by: Joerg Bornemann <joerg.bornemann@nokia.com>
we just determined that the file does not exist, so it's entirely
pointless to query its type from the file system. consequently, the
respective fallback branch would assume a regular file anyway.
Change-Id: I42590ffc2a5f650fb430a9398cb1859217ed4350
Reviewed-by: Joerg Bornemann <joerg.bornemann@nokia.com>
the value is still re-processed numerous times, end each "exit path"
does own escaping, while not every path can deal with an escaped path.
Change-Id: I0bf4a043809bf4b7877d02e5d8dfe8f794a7dd00
Reviewed-by: Joerg Bornemann <joerg.bornemann@nokia.com>
Remove references to an old compiler that has not been
supported for a long time. Also remove Borland specific
configuration flags which have no meaning elsewhere.
Change-Id: I3634a52b78f737ea972073e14c2b6669dcd0ae63
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@nokia.com>
the visitor-pattern like approach is not needed any more
Change-Id: I990db681cbeee91d89ecba97745a8104595247e7
Reviewed-by: Friedemann Kleint <Friedemann.Kleint@nokia.com>
Reviewed-by: Joerg Bornemann <joerg.bornemann@nokia.com>
it's more elegant, and more similar code is better.
Change-Id: I2b8b036cb70a932fd171e23cf7d3389188401924
Reviewed-by: Friedemann Kleint <Friedemann.Kleint@nokia.com>
Reviewed-by: Joerg Bornemann <joerg.bornemann@nokia.com>
the unix and windows variants don't rely on it.
not making it purely virtual only because the project generator inherits
MakefileGenerator as well but does not need an implementation.
Change-Id: I80099b3f5d07cd037b408cf1099c58ff3a2904cd
Reviewed-by: Friedemann Kleint <Friedemann.Kleint@nokia.com>
Reviewed-by: Joerg Bornemann <joerg.bornemann@nokia.com>
this way QMAKE_LIBS{,_PRIVATE} can be treated the same as in all other
generators, which allows us to:
- make the windows generators' findLibraries() be more like the unix
version
- dispose of QMAKE_INTERNAL_PRL_LIBS handling while reading prl files
(because the output always goes to QMAKE_LIBS)
- as a side effect, QMAKE_LIBS_PRIVATE are not subjected to prl file
resolution any more, which is again consistent with unix - the
correctness of that needs to be assessed separately.
Change-Id: Ie9bc04d117eff6a7cde846677f98acf3c64aa6ee
Reviewed-by: Friedemann Kleint <Friedemann.Kleint@nokia.com>
it's a tad insane to expect the user to do that
Change-Id: I75c68f2a28656c9ba2e3fabcc79718b899b29ce7
Reviewed-by: Joerg Bornemann <joerg.bornemann@nokia.com>
Don't hardcode the "qt_config" EXTRA variable and use QMAKE_PKGCONFIG_VARIABLES instead.
This allows qmake create the .pc files that are unrelated to Qt.
Change-Id: Ic72005e8819a15f6c50f3aaf79424a247fba20af
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@nokia.com>
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>
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>
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>
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>
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>
instead of making the "real" targets depend on the makefiles, add
conditional makefile generation to the targets themselves.
this causes makefile generation to follow the recursion order determined
by the project, which is important when dealing with prl and module pri
files.
a side effect of this is that qmake and make calls are interleaved now,
which is entirely different from a 'qmake -r' run.
on the downside, calling make with multiple targets which operate on the
same subprojects without prior makefile generation will make a mess, as
the qmake calls will be racing. this should be no problem, as qmake does
not generate recursive targets where this would be useful - at least by
default.
it is not sufficient to just order the creation of the makefiles
non-recursively (e.g., by using gnu-specific order-only-prerequisites),
as an interrupted and subsequently resumed build would happily skip the
nested makefiles.
workable alternative approaches would be walking the entire tree in a
pre-pass to ensure makefile presence (which is incredibly slow) or
creating additional stamp files only after recursing and having the
makefiles depend on them (which is ugly).
Task-number: QTBUG-23376
Reviewed-by: Joerg Bornemann <joerg.bornemann@nokia.com>
Change-Id: I88d3e7610215677d362026de316513d3bea04b06
instead of hard-coding platform differences, use a variable.
Change-Id: I20e98811ad5f07429148c6f88aedbabc3ba58fff
Reviewed-by: Joerg Bornemann <joerg.bornemann@nokia.com>
everything in the projects should be normalized. only the makefile
generators need to adjust it to the native form.
Change-Id: I06a4e997f32134d13949ec4a9dd1b44367aab7cb
Reviewed-by: Joerg Bornemann <joerg.bornemann@nokia.com>
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@nokia.com>
no point in entering the outdir and leaving the indir.
this doesn't really matter except for the obsolete borland generator,
but whatever.
Change-Id: Ieb4e4c549ebd65f99e4d00819c52822968b060a4
Reviewed-by: Joerg Bornemann <joerg.bornemann@nokia.com>
the code above already deals with the differences
Change-Id: Ifb799e46f5187e7bd3d0f0169e868ad267bcfe23
Reviewed-by: Joerg Bornemann <joerg.bornemann@nokia.com>
no need to define an obscure variable for it. just inline it.
the assignments are left in for compatibility with hand-written commands.
Change-Id: I9bc3914e2c4116f3b8fe00a421ca0f036bb7e214
Reviewed-by: Joerg Bornemann <joerg.bornemann@nokia.com>
so far it would rebuild the makefiles one level down only, which is
somewhat arbitrary and not really helpful.
Change-Id: I5fe01f379ecc4b210610a674d7df7dfc18131eef
Reviewed-by: Joerg Bornemann <joerg.bornemann@nokia.com>
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@nokia.com>
that's consistent with other projects. 'qmake_all' can be used for
recursion.
Change-Id: Ie6d620f7a3e0e28d3f2f82f01ca94c2f46137c68
Reviewed-by: Joerg Bornemann <joerg.bornemann@nokia.com>
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@nokia.com>
$(MAKEFILE) is our own file, and using it would wreak havoc if it was
named anything but Makefile.
Change-Id: I51cae2014a85399b409f18788f864ff2a82e493e
Reviewed-by: Joerg Bornemann <joerg.bornemann@nokia.com>
this is the source dir equivalent of .qmake.cache and can be checked
into the repository. it can be used to make project-wide settings, or
just to signal the presence of an mkspecs/ directory.
Change-Id: I5f1cebad2aa2a89e78089341b2051613b6b7a613
Reviewed-by: Joerg Bornemann <joerg.bornemann@nokia.com>
this allows the creation of aggregator projects, like, say, qt5.
this is not expected to have a negative impact, as no project could
reasonably expect a nested .qmake.cache to *not* take effect - in fact,
if the project was processed stand-alone, it would already use it.
Change-Id: I33f2935d309baba7e95465f2fefb8231c4f03eda
Reviewed-by: Joerg Bornemann <joerg.bornemann@nokia.com>
properties are now split into a write location $$[FOO] and a read
location $$[FOO/get]. the write locations are hard-coded and configurable
via qt.conf/Paths as before, while the read locations are configured via
qt.conf/EffectivePaths.
this finally provides a clean solution to the problem that during the qt
build itself tools and libraries need to be taken from somewhere else
than they are installed to.
Change-Id: I956c43bd082afd465e690fe75d0bee3c2c0f7c25
Reviewed-by: Joerg Bornemann <joerg.bornemann@nokia.com>
Reviewed-by: Robin Burchell <robin+qt@viroteck.net>
make_first really means the same, as the default target is 'first'
which immediately invokes make_first.
as a nice side effect, this fixes ALL_DEPS not being used by the subdirs
template's default target - make_default wasn't special-cased in the
right place, while make_first is.
Change-Id: I946cea3b2dae198eda61d31dbd19c6ccd15deac7
Reviewed-by: Marius Storm-Olsen <marius.storm-olsen@nokia.com>
when qmake runs into the new option(host_build) command, it will restart
the project evaluation with a host spec.
the new default host spec is called default-host (gasp!). it is
overridden with the pre-exising -spec / -platform option, while the new
-xspec / -xplatform option overrides the pre-existing default spec.
specifying -spec but not -xspec will set the xspec, too, so the behavior
is backwards-compatible. same for the XQMAKESPEC override read from
.qmake.cache and the environment variable.
the cleaner solution would be adding -hostspec, to be symmetrical with
the override semantics, but that would deviate from configure in turn.
Change-Id: I4297c873780af16ab7928421b434ce0f1d3820da
Reviewed-by: Joerg Bornemann <joerg.bornemann@nokia.com>
219a688ac1 added the version extension
which only occurs on the DLL import library.
Change-Id: Ie913adf039d7f3a9aadf43f4af014289a8aeacbe
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@nokia.com>
Reviewed-by: Hib Eris <hib@hiberis.nl>
this was a somewhat magic support for sysroots, automatically rewriting
a number of path-holding variables. this was (as usual) completely
undocumented, extremely fragile, and we are coming up with something
better now anyway.
Change-Id: I045910f532cb3efc839ea81c7a48f8db695e4092
Reviewed-by: Joerg Bornemann <joerg.bornemann@nokia.com>
merge them into before_user_vars. they are evaluated right after another
anyway.
Change-Id: I11859284b363fee01233f6e20989444fef711d0d
Reviewed-by: Joerg Bornemann <joerg.bornemann@nokia.com>
Reviewed-by: Marius Storm-Olsen <marius.storm-olsen@nokia.com>
instead of being a variable added to the makespec (via qconfig.pri),
QT_SYSROOT is now a property.
the QT_INSTALL_... properties are now automatically prefixed with the
sysroot; the raw values are available as QT_RAW_INSTALL_... - this is
expected to cause the least migration effort for existing projects.
-hostprefix and the new -hostbindir & -hostdatadir now feed the new
QT_HOST_... properties.
adapted the qmake feature files and the qtbase build system accordingly.
Change-Id: Iaa9b65bc10d9fe9c4988d620c70a8ce72177f8d4
Reviewed-by: Marius Storm-Olsen <marius.storm-olsen@nokia.com>
The libName that is used in pkgConfig files should include the
TARGET_VERSION_EXT.
This is needed because in Win32MakefileGenerator::fixTargetExt(),
the TARGET_VERSION_EXT is added to the library name.
In Win32MakefileGenerator::processVars(), if TARGET_VERSION_EXT is empty,
it is set to VER_MAJ.
On platforms != Windows, TARGET_VERSION_EXT does not seem to be used.
We probably got away with this so far because pkgconfig files generation
for win32 was just added in 4.8 and nobody uses them yet, and because on
platforms != windows the TARGET_VERSION_EXT is not used.
Change-Id: I56f239e389f0ef926030e4c2376cadd92c4f673c
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@nokia.com>
It was the result of miscoordination between Andreas Holzammer and me,
duplicating the functional change of commit
ea2c9f764f.
This reverts commit 8b7a9b4898.
Change-Id: I9a47746c1c12ca00b2dc5c5d50e99f9bf990e3f7
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@nokia.com>
qmake-generated Makefiles since 4.6 (according to bug report) contain
recursive qmake invocations with absolute paths to .pro files, where the
absolute paths are not escaped.
Task-number: QTBUG-11776
Change-Id: I49772c7d9147efea3bb7b9623c61d92aa0bb6857
Reviewed-by: Jonas Gastal <jgastal@profusion.mobi>
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@nokia.com>
As in the past, to avoid rewriting various autotests that contain
line-number information, an extra blank line has been inserted at the
end of the license text to ensure that this commit does not change the
total number of lines in the license header.
Change-Id: I311e001373776812699d6efc045b5f742890c689
Reviewed-by: Rohan McGovern <rohan.mcgovern@nokia.com>
When installing extra binaries the paths where not being escaped,
causing stripping and deleting to fail when dealing with filenames
that contained one or more spaces.
Change-Id: Iba4517b1110f6af30f7e2662cb86024a8b7b81f7
Reviewed-by: Kurt Korbatits <kurt.korbatits@nokia.com>
Reviewed-by: Lincoln Ramsay <lincoln.ramsay@nokia.com>
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@nokia.com>
this should make the evaluator quite a lot faster. the total win for
qtbase/src is only 6%, though.
i made some effort to avoid that output files get randomized. however, i
didn't bother to keep debug output sorted.
Change-Id: Id9cef4674c0153c11ebbb65cb63bf8c229eb56e3
Reviewed-by: Joerg Bornemann <joerg.bornemann@nokia.com>
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@nokia.com>
This solves the problem of two source files in the same project
having the same name, which ends up with colliding object files.
It also solves issues with moving files around in the source tree
when the GNUmake option is used together with gcc_MD_depends,
where the dependency file would end up with a dependency to the
old source location.
We already have object_with_source option, but that does not play
nicely with shadow-builds. The new option tries to keep the same
hirarcy in the output dir (while still respecting OBJECTS_DIR),
as the sources. Any source with a path outside of the root project
directory (absolute or relative, ../../foo.cpp), will be put inside
the objects directory with the full path appended.
Change-Id: I0c96291974f4801f4c55a26485d512b15c144c00
Reviewed-by: Marius Storm-Olsen <marius.storm-olsen@nokia.com>
Extra compilers may define a depend_command that's used to generate
dependencies for each input. When GNUmake was enabled we failed to
run this command, which was affecting resource files, as resource
dependencies are handled by an extra compiler defined in resources.prf.
The result was that changes to resources included in a resources-file
did not trigger a re-run of qrc and subsequent recompile of the
resource object file.
We must always run these custom dependency commands, even when GNUmake
(and the extended gcc_MD_depends option) is enabled, as GCC is only able
to handle regular #include-type dependencies. Hence, the check for the
'include_deps' flag was removed from doDepends(), and the check for
GNUmake was moved to the one place where it still made sense -- when
deciding whether or not to do recursive dependency checking.
Change-Id: I5ddb75c873120c90f798808efc52e81500786301
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@nokia.com>
Reviewed-by: Simon Hausmann <simon.hausmann@nokia.com>
Reviewed-by: Marius Storm-Olsen <marius.storm-olsen@nokia.com>
And enable this configuration option for the resource compiler. This
results in a re-run of qmake whenever you touch a qrc file, which is
needed to keep the dependencies up to date. Otherwise you might end
up in the situation where you add a file to a qrc, edit the file some
time later, but a rebuild does not regenerate a cpp file and compile
that, so the final binary is stale.
Technically this dependency problem is present for all source files,
and qrc files are no different than any cpp file that you add a new
header #include to, or adding a Q_OBJECT macro to a header. To pick
up these changes we have to re-run qmake, so that qmake can run its
internal dependency checking, and any extra compiler dependency
commands.
The reason we're making this change for rcc files it that conceptually
people treat them as a "project" files, and expect them to behave similarly
to .pro or .pri files, in that editing the file will invalidate the
makefile. In practice this is often what happens when adding new
headers, as you touch the project file when changing the HEADERS
variable.
Task-number: QTBUG-13334
Change-Id: If69149678e7fba6d812d31dcc17877427f9a6122
Reviewed-by: Simon Hausmann <simon.hausmann@nokia.com>
Reviewed-by: Kai Koehne <kai.koehne@nokia.com>
Recursive QMAKE_EXTRA_TARGETS were omitting the `-f' option to make.
This would break in the case where the correct makefile was not named
`Makefile'. The included autotest demonstrates the problem.
Note that this was fixed for normal targets back in 2005
by faac7bd178654fd67a6f3f9cf4f6f2605071448d (p4 202370), but was not
fixed for extra targets.
Reviewed-by: ossi
(cherry picked from commit 96a3bf7a8bbc1e5361e16cbeeceb4be674b88c30)
This is the beginning of revision history for this module. If you
want to look at revision history older than this, please refer to the
Qt Git wiki for how to use Git history grafting. At the time of
writing, this wiki is located here:
http://qt.gitorious.org/qt/pages/GitIntroductionWithQt
If you have already performed the grafting and you don't see any
history beyond this commit, try running "git log" with the "--follow"
argument.
Branched from the monolithic repo, Qt master branch, at commit
896db169ea224deb96c59ce8af800d019de63f12
