| Commit message (Collapse) | Author | Age | Files | Lines |
| |
|
| |
|
| |
|
| |
|
| |
|
|
|
|
|
| |
Decrement reference to temporary string object only after last use of its
value.
|
| |
|
|
|
|
|
|
|
|
|
|
|
| |
Formerly, trying to build gi/Foo-1.0.gir resulted in namespace being set to "gi/Foo"
Trying to build g-i/Foo-1.0.gir was even setting it to "g"
This fixes this behaviour by only considerating the filename without dirs.
Btw, ensure that the directory exists
Signed-off-by: Marc-Antoine Perennou <Marc-Antoine@Perennou.com>
https://bugzilla.gnome.org/show_bug.cgi?id=665276
|
| |
|
|
|
|
|
|
| |
use TOOL_SUBSTITUTIONS like the others, to fix an unsubstitued @datarootdir@ in g-ir-doc-tool
https://bugzilla.gnome.org/show_bug.cgi?id=659649
|
|
|
|
|
|
| |
These variables are unused.
https://bugzilla.gnome.org/show_bug.cgi?id=665249
|
|
|
|
|
|
|
| |
g_slice_free complains when you pass it a pointer of a different type
than it was expecting, like it should.
https://bugzilla.gnome.org/show_bug.cgi?id=665249
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
That is how gcc expects them.
From the gcc man page:
-llibrary
Search the library named library when linking.
It makes a difference where in the command you write this option; the linker
searches and processes libraries and object files in the order they are
specified. Thus, foo.o -lz bar.o searches library z after file foo.o but
before bar.o. If bar.o refers to functions in z, those functions may not be
loaded.
https://bugzilla.gnome.org/show_bug.cgi?id=664616
|
|
|
|
| |
https://bugzilla.gnome.org/show_bug.cgi?id=657202
|
|
|
|
|
|
|
| |
Gio no longer draws in gmodule, so we need to add the dependency
explicitly.
https://bugzilla.gnome.org/show_bug.cgi?id=664681
|
|
|
|
| |
https://bugzilla.gnome.org/show_bug.cgi?id=664681
|
| |
|
|
|
|
|
|
| |
situations. So adjust resolve_non_libtool() in this case.
https://bugzilla.gnome.org/show_bug.cgi?id=664282
|
|
|
|
|
| |
This is needed so that g-ir-scanner can find the libxml2 include
path when producing bindings for libraries that interact with libxml2.
|
|
|
|
|
| |
We shouldn't do it for module-internal paths, since it can't work
and is unnecessary even if it did.
|
|
|
|
|
|
|
|
|
|
|
| |
Introspection scanning is a bit special in that we both compile *and*
run a temporary binary. In some cases like jhbuild we will pick up
the runtime location of libraries via LD_LIBRARY_PATH. But this falls
over when we've been given a private subdirectory, since it won't be
in that path.
Let's just "do the right thing" here by injecting an rpath for the
directories as well.
|
| |
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Fix return type annotation for gi_marshalling_tests_array_gvariant_in()
Add function variations of gi_marshalling_tests_array_gvariant_i() with
different transfer modes.
Make gi_marshalling_tests_array_gvariant_none_in() actually respect
the transfer none annotation on the return value.
https://bugzilla.gnome.org/show_bug.cgi?id=638915
Signed-off-by: Martin Pitt <martin.pitt@ubuntu.com>
|
|
|
|
|
| |
This is a requisite for writing a pygobject test case for
https://bugzilla.gnome.org/show_bug.cgi?id=662383
|
|
|
|
| |
Spotted by Rico Tzschichholz
|
|
|
|
|
|
| |
By simply appending '.dll' to the library names.
https://bugzilla.gnome.org/show_bug.cgi?id=620566
|
| |
|
|
|
|
|
|
|
| |
A GIBaseInfo struct can underneath either be GIRealInfo *or*
GIUnresolvedInfo if the type is GI_INFO_TYPE_UNRESOLVED. So when we
eventually free the structures slice use the correct struct type
otherwise things get unhappy.
|
| |
|
|
|
|
|
| |
This is useful for someone who just wants a set of constants from a .h
file accessible by introspection.
|
|
|
|
|
|
|
| |
We don't need to, and if we wanted to, we have to link against
-lgthread.
Spotted by Rico Tzschichholz
|
|
|
|
|
|
|
|
|
|
| |
error was not set.
Noticed the segmentation fault while using Vala to generate a .gir, a bug has
been filed tomake sure Vala doesn't export gir symbols outside of a namespace
(see https://bugzilla.gnome.org/show_bug.cgi?id=661952)
https://bugzilla.gnome.org/show_bug.cgi?id=661951
|
|
|
|
| |
This helps debug. Also, add a 'fatal' debug break.
|
| |
|
|
|
|
|
|
|
| |
6172c268 made the scanner think GObject* wasn't a superclass of its
subclasses. Fix that.
https://bugzilla.gnome.org/show_bug.cgi?id=661677
|
|
|
|
| |
g_thread_init() is deprecated, and unnecessary since GLib 2.24.
|
|
|
|
|
|
|
|
| |
This fixes compilation where CC="ccache gcc" and similar.
https://bugzilla.gnome.org/show_bug.cgi?id=660160
Signed-off-by: Natanael Copa <ncopa@alpinelinux.org>
|
|
|
|
| |
https://bugzilla.gnome.org/show_bug.cgi?id=654069
|
|
|
|
|
|
|
|
| |
This reverts commit e17c09749c23edbd928db19a8162484ddb6299b8.
We actually want the test for sizeof(GValue).
https://bugzilla.gnome.org/show_bug.cgi?id=660338
|
|
|
|
| |
https://bugzilla.gnome.org/show_bug.cgi?id=660523
|
|
|
|
|
|
|
| |
It's redundant since we're already checking that it's equal to GValue,
and a problem is that padding may break exact assertions on the size.
https://bugzilla.gnome.org/show_bug.cgi?id=660338
|
|
|
|
|
|
|
| |
On some platform like armel or powerpc the compiler adds extra padding to
structs. Make the test_size_of_struct_with_array_of_anon_unions() check more
liberal by only requiring that the size of the struct is at least as big as
expected. Also use g_assert_cmpuint() to make it easier to see the actual diff.
|
|
|
|
|
|
| |
Otherwise we get warnings for params that have been skipped
https://bugzilla.gnome.org/show_bug.cgi?id=660352
|
|
|
|
|
|
| |
It was returning the wrong data.
https://bugzilla.gnome.org/show_bug.cgi?id=659749
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
I set ext to None in commit f2a70843905839a1b11795310d2aa3b85a474e77
without realizing _generate_tempfile does a simple string formatting.
A simple test proves this does not generate an empty string:
$ python -c "f = '%s' % None; print f, type(f)"
None <type 'str'>
So pass an empty string instead of None for the suffix parameter.
https://bugzilla.gnome.org/show_bug.cgi?id=659502
|
|
|
|
|
|
|
|
|
|
|
|
| |
- AM_CHECK_PYTHON_LIBS doesn't work for 64bit sytems (lib64)
- Python extension modules do not need to be linked against
libpython on linux anyway, but it is needed on Windows
So only run AM_CHECK_PYTHON_LIBS in configure.ac for the windows
case and make it clear in Makefile-giscanner.am linking
against libpython is a windows only thing.
https://bugzilla.gnome.org/show_bug.cgi?id=658914
|
|
|
|
|
|
| |
* for now we assume simple struct arrays and gvalue arrays are flat
until gi gets support for annotating the level of indirection
of an array.
|
|
|
|
|
| |
Add an annotation tag "Value:" which can be used on
constants to override the value.
|