As pointed out by gtk-doc, these are all symbols which have been marked
as deprecated, but which aren’t protected by a deprecation guard. We
can’t use G_DEPRECATED_IN_* for them, as they are all non-function
symbols. Instead, wrap them in #ifndef G_DISABLE_DEPRECATED.
In some cases, we also need to wrap one or two functions which use the
deprecated types in G_DISABLE_DEPRECATED too.
Signed-off-by: Philip Withnall <withnall@endlessm.com>
In order to allow GLib itself to be built with G_DISABLE_ASSERT defined,
we need to explicitly undefine it when building the tests, otherwise
g_test_init() turns into an abort.
Signed-off-by: Philip Withnall <withnall@endlessm.com>
Fixes: #1708
g_object_set_data() should only ever be used with a small, bounded set
of keys, or the memory usage of the quark lookup table will grow
unbounded. Document that.
Signed-off-by: Philip Withnall <withnall@endlessm.com>
Fixes: #682
Grouping things together makes them easier to find and keep up to date.
This doesn’t modify any of the comments or make any functional changes.
Signed-off-by: Philip Withnall <withnall@endlessm.com>
They were originally generated by glib-genmarshal, which documents its
output as being under the same license as the containing project. In
this case, that’s LGPL.
Signed-off-by: Philip Withnall <withnall@endlessm.com>
The threads used to iterate at least 10000 times before setting the
"seen thread" flag to true. After porting they inadvertently did that
in the first iteration.
Previously, the test assumed that thread1 and thread2 would be scheduled
enough to set seen_thread{1,2} by the fact that the test runs for a high
number of iterations. On some platforms/schedulers, that’s not true,
which causes the test to spuriously fail.
Fix that by forcing the test to continue iterating until both threads
are seen. If this takes too long, the Meson test runner timeout will be
hit and the test will be terminated.
Signed-off-by: Philip Withnall <withnall@endlessm.com>
Previously, all three threads would access several global variables
without locking.
Fix that by using atomic accesses to data stored within the
test_closure_refcount() function, which also eliminates the global state
(which would confuse further tests if they were added to this file).
Signed-off-by: Philip Withnall <withnall@endlessm.com>
g_assert() can be compiled out if G_DISABLE_ASSERT is defined; and
g_assert_*() provide more specific error messages on failure.
Signed-off-by: Philip Withnall <withnall@endlessm.com>
These functions are not run more than once, so the variables don’t need
to be static to save state between runs.
Signed-off-by: Philip Withnall <withnall@endlessm.com>
GValue g_type field is used for synchronization with g_once_init_enter,
and so it should be written to only with g_once_init_leave.
Replace structure copy with memcpy that copies the one remaining field
of GValue, i.e., data array.
It's necessary sometimes for installed tests to be able to run with a
custom environment. For example, the gsocketclient-slow test requires an
LD_PRELOADed library to provide a slow connect() (this is to be added in
a followup commit).
Introduce a variable `@env@` into the installed test template, which we
can override as necessary when generating `.test` files, to run tests
prefixed with `/usr/bin/env <LIST OF VARIABLES>`.
As the only test that requires this currently lives in `gio/tests/`, we
are only hooking this up for that directory right now. If other tests in
future require this treatment, then the support can be extended at that
point.
I'm trying to use `-fsanitize=thread` for OSTree, and some of
these issues seem to go into GLib. Also, the sanitizers work better if
the userspace libraries are built with them too.
This fix is similar to
b6814bb37c
Mixing atomic and non-atomic reads trips TSAN, so let's change the
assertions to operate on the local values returned from atomic
read/writes.
Without this change I couldn't even *build* GLib with TSAN, since we
use gresources during compilation, which uses GSubprocess, which hits
this code.
(Minor review fixes made by Philip Withnall <withnall@endlessm.com>.)
https://gitlab.gnome.org/GNOME/glib/issues/1224
This is an analogous commit to c1f5e528. The original fix only touched
gtype.h and not gtypemodule.h.
The *_init() functions have prototypes incompatible with *InitFunc types they
are being cast to. This upsets GCC 8's -Wcast-function-type that's enabled by
default with -Wextra.
Let's not have the public header files emit a warning and neutralize it by
doing a void(*)(void) cast first.
Signed-off-by: Philip Withnall <withnall@endlessm.com>
https://gitlab.gnome.org/GNOME/glib/issues/1666
So long, and thanks for everything. We’re a Meson-only shop now.
glib-2-58 will remain the last stable GLib release series which is
buildable using autotools.
We continue to install autoconf macros for autotools-using projects
which depend on GLib; they are stable API.
Signed-off-by: Philip Withnall <withnall@endlessm.com>
And mention why it’s not a GInterfaceInitFunc as people who have read
the GObject docs cover-to-cover might expect.
Signed-off-by: Philip Withnall <withnall@endlessm.com>
Note that it's not reported with gcc. It's only reported with g++.
C++ code to reproduce this warning:
#include <glib-object.h>
G_BEGIN_DECLS
#define GARROW_TYPE_FILE (garrow_file_get_type())
G_DECLARE_INTERFACE(GArrowFile,
garrow_file,
GARROW,
FILE,
GObject)
struct _GArrowFileInterface {
GTypeInterface g_iface;
};
G_DEFINE_INTERFACE(GArrowFile,
garrow_file,
G_TYPE_OBJECT)
static void
garrow_file_default_init(GArrowFileInterface *iface)
{
}
G_END_DECLS
Build command line:
% g++ -Wall -shared -o liba.so a.cpp $(pkg-config --cflags --libs gobject-2.0)
Message:
In file included from /tmp/local.glib/include/glib-2.0/gobject/gobject.h:24,
from /tmp/local.glib/include/glib-2.0/gobject/gbinding.h:29,
from /tmp/local.glib/include/glib-2.0/glib-object.h:23,
from a.cpp:1:
a.cpp: In function 'GType garrow_file_get_type()':
/tmp/local.glib/include/glib-2.0/gobject/gtype.h:219:50: warning: '<<' in boolean context, did you mean '<' ? [-Wint-in-bool-context]
#define G_TYPE_MAKE_FUNDAMENTAL(x) ((GType) ((x) << G_TYPE_FUNDAMENTAL_SHIFT))
~~~~~^~~~~~~~~~~~~~~~~~~~~~~~~~~~
/tmp/local.glib/include/glib-2.0/gobject/gtype.h:2026:11: note: in definition of macro '_G_DEFINE_INTERFACE_EXTENDED_BEGIN'
if (TYPE_PREREQ) \
^~~~~~~~~~~
/tmp/local.glib/include/glib-2.0/gobject/gtype.h:1758:47: note: in expansion of macro 'G_DEFINE_INTERFACE_WITH_CODE'
#define G_DEFINE_INTERFACE(TN, t_n, T_P) G_DEFINE_INTERFACE_WITH_CODE(TN, t_n, T_P, ;)
^~~~~~~~~~~~~~~~~~~~~~~~~~~~
a.cpp:16:1: note: in expansion of macro 'G_DEFINE_INTERFACE'
G_DEFINE_INTERFACE(GArrowFile,
^~~~~~~~~~~~~~~~~~
/tmp/local.glib/include/glib-2.0/gobject/gtype.h:178:25: note: in expansion of macro 'G_TYPE_MAKE_FUNDAMENTAL'
#define G_TYPE_OBJECT G_TYPE_MAKE_FUNDAMENTAL (20)
^~~~~~~~~~~~~~~~~~~~~~~
a.cpp:18:20: note: in expansion of macro 'G_TYPE_OBJECT'
G_TYPE_OBJECT)
^~~~~~~~~~~~~
Conceptually the binding is kept alive as long as both the source and
target exist. This means that an API user needs to take some care to
either hold a reference or only use a pointer to the binding as long as
also holding references to both objects.
Clarify the documentation a bit.
Turns out the fix in commit 93555577c wasn't enough, when using glib as
subproject and the parent project uses only libgio_dep, and include
<gi18n.h>, it won't find libintl.h because it's in the
include_directories of libglib_dep. Fix that by declaring dependencies
explicitly, which is the right thing to do since glib and gobject are
public dependencies of gio. That reflects what we do for the pkg-config
file as well.
When passing a function to G_IMPLEMENT_INTERFACE, it actually has to
take two arguments. Who knew?
Signed-off-by: Philip Withnall <withnall@endlessm.com>
When using glib as subproject we are forced to pass glib_dep,
gobject_dep and gio_dep to any build target. If we pass only gio_dep it
will missing include directory for glib and gobject.
Add tests using an object declared with G_DECLARE_FINAL_TYPE, that is derived
from another, declared using G_DECLARE_DERIVABLE_TYPE, and that
thus uses _GLIB_DEFINE_AUTOPTR_CHAINUP to define cleanup functions.
And verify that both g_autoptr(Type) and g_auto(s)list(Type) work
Fedora is using https://fedoraproject.org/wiki/Changes/Annobin
to try to ensure that all objects are built with hardening flags.
Pass down `CFLAGS` to ensure the SystemTap objects use them.
An assertion is harder to skip over, and using a g_critical() can give
us a more informative error message.
Signed-off-by: Philip Withnall <withnall@endlessm.com>
https://gitlab.gnome.org/GNOME/glib/issues/179
This makes it easier to debug test failures, by ensuring that g_debug()
and g_test_message() are printed as TAP diagnostics.
Fixes: https://gitlab.gnome.org/GNOME/glib/issues/1528
Signed-off-by: Simon McVittie <smcv@collabora.com>
g_object_bind_property() (transfer none) returns a GBinding with an existing internal
reference which is active as long as the "binding" is. This allows to optionally use
the binding without any memory management, as it will remove itself when it is no longer
needed.
There are currently three ways to remove the "binding" and as a result the reference:
1) Either the source or target dies and we get notified by a weakref callback
2) The user unrefs the binding until it is destroyed (which is semi-legal,
but worked and is used in the test suite)
3) The user calls g_binding_unbind()
In case (3) the problem was that it always calls unref even if the "binding" is already
gone, leading to crashes when called from bindings multiple times.
In #1373 and !197 it was noticed that a function always unrefs which would be a
"transfer full" annotation, but the problem here is that it should only remove the
ref when removing the "binding" and the annotation should stay "transfer none".
As a side effect of this fix it is now also possible to call g_binding_unbind() multiple
times where every call after the first is a no-op.
This also adds explicit tests for case (1) and (3) - only case (3) is affected by this change.
Part of runMkenumsWithHeader() was duplicated in test_reproducible(),
and would otherwise need to be duplicated again in upcoming tests. Many
places duplicated decoding stdout/stderr and checking the exit code.
Introduce a named tuple for the returned fields; and factor out writing
a template file to pass with --template.
The new python module, added with 0.46, works with Python 2 and 3 and
allows to pass a path for the interpreter to use, if the need arises.
Previously the meson build set PYTHON, used in the shebang line of
the scripts installed by glib, to the full path of the interpreter.
The new meson module doesn't expose that atm, but we should set it to
a executable name anyway, and not a full path.
Several of our tools are installed and are used by other projects to
generate code. However, there is no 'install' when projects use glib
as a subproject.
We need some way for glib to 'provide' these tools so that when some
project uses glib as a subproject, find_program('glib-mkenums') will
transparently return the glib-mkenums we just built.
Starting from Meson 0.46, this can be done with the
`meson.override_find_program()` function.
As a bonus, the Meson GNOME module will also use these
'overriden'/'provided' programs instead of looking for them in PATH.
PEP8 says that:
"Comparisons to singletons like None should always be done with is or
is not, never the equality operators."
glib uses a mix of "== None" and "is None". This patch changes all
cases to the latter.
The implementation is silently discarding this anyway, and
g_object_unref() is using atomic operations. So this should be safe.
Having this here triggers -Wdiscarded-qualifiers when g_clear_pointer()
is fixed to use __typeof__().
The implementation is silently discarding this anyway, and
g_object_unref() is using atomic operations. So this should be safe.
Having this here triggers -Wdiscarded-qualifiers when g_clear_pointer()
is fixed to use __typeof__().
Switch the check which tests whether the object has been finalised from
being a use-after-free, to using a weak pointer which is nullified on
finalisation.
Signed-off-by: Philip Withnall <withnall@endlessm.com>
If some other per value option was present than 'skip' or 'nick' then
a KeyError would occur. Ignoring such options matches the behaviour of
the old, Perl-based glib-mkenums.
Signed-off-by: Peter Kjellerstedt <peter.kjellerstedt@axis.com>
https://gitlab.gnome.org/GNOME/glib/issues/1360
This makes them a bit more unique (and, crucially, in the g_* namespace)
to avoid shadowing collisions with calling code.
Signed-off-by: Philip Withnall <withnall@endlessm.com>
https://gitlab.gnome.org/GNOME/glib/issues/258
It's been 4 years and 8 development cycles since we introduced
G_ADD_PRIVATE and offset-based private data access. It is now
time to finally deprecate the old mechanism.
Closes: #699
Meson has the ability to classify tests according to "suites", a list of
tags. This is especially useful when we want to run specific sets of
tests — e.g. only GLib's tests — instead of the whole test suite. It
also allows us to classify special tests, like "slow" ones, so that we
can only run them when needed.
In master, it is already possible to build GLib using Visual Studio
using Meson[1] for some time, so we should focus on maintaining only the
Meson build files for building GLib with Visual Studio.
[1]: There are caveats when building with Visual Studio 2008, namely
that one needs to use the mt command to embed the manifests that
are generated with the .exe/DLLs, for all builds, and that in the
case where the compilation hangs on Visual Studio 2008 x64, as a
workaround, should stop the build by terminating all cl.exe tasks
and change the compiler optimization flag from /O2 (full speed) to
/O1 (optimize for size), due to compiler optimization issues.
It's mostly not used anymore and doesn't do what it says it does.
The docs state that it affects GList, GSList, GNode, GMemChunks, GSignal,
GType n_preallocs and GBSearchArray while:
* GList, GSList and GNode use GSlice and are not affected
* GMemChunks is gone
* GType npreallocs is ignored
It also states that it can be used to force the usage of g_malloc/g_free,
which is handled by G_SLICE=always-malloc now.
The only places where it's used is in signal handling through GBSearchArray
and in GValueArray (deprecated). Since it's unlikely that anyone wants to
reduce allocation sizes just for those cases remove the build option.
valgrind.h is a verbatim copy taken from Valgrind project. Previously
that file had local changes that got dropped by last update. To avoid
regressing again, do not edit valgrind.h anymore and instead add a
gvalgrind.h wrapper that gets included instead.
This fix 2 errors:
- uintptr_t is not defined when including valgrind.h on mingw.
- MSVC compiler is not supported on amd64-Win64 platform.
The timer tests expect that a small value for sleep does not result in
no sleep at all. Round up to the next millisecond to bring it more in line
with other platforms.
This fixes the glib/timer tests.
This makes the 'threadtests' time out since that uses small usleeps a lot and
until now didn't wait at all, but now always waits a msec. Reduce the amount
of tests done on Windows to get the runtime down to something reasonable again.
https://bugzilla.gnome.org/show_bug.cgi?id=795569
On non-glibc platforms gettext is provided by extra libintl dependency.
We wrongly thought libintl is an internal dependency and applications
needs to explicitly link on it, but turns out that breaks many
applications and with autotools the .pc generated actually has -lintl in
public "Libs:".
https://bugzilla.gnome.org/show_bug.cgi?id=796085
Do not add support for its subtypes, since all their constructors return
GParamSpec*, and g_param_spec_unref() takes a GParamSpec* rather than a
gpointer — adding G_DEFINE_AUTOPTR_CLEANUP_FUNC() for subtypes of
GParamSpec results in compiler warnings about mismatched parameter
types (GParamSpecBoolean* vs GParamSpec*, for example).
Signed-off-by: Philip Withnall <withnall@endlessm.com>
https://bugzilla.gnome.org/show_bug.cgi?id=796139
-z nodelete breaks the libresourceplugin module usage in the resources.c
test, which expects to be able to unload it.
Make the Meson build match what the autotools build does: only pass
glib_link_flags to the headline libraries (glib-2.0, gio-2.0,
gobject-2.0, gthread-2.0, gmodule-2.0) and omit it from all other build
targets.
https://bugzilla.gnome.org/show_bug.cgi?id=788771
The -fstack-protector-strong used in many distributions by default has a
rather drastic slowdown of the fast path in generated _get_type()
functions using G_DEFINE_* macros. The amount can vary by architecture,
GCC version, and compiler flags.
To work around this, and ensure a higher probability that our fast-path
will match what we had previously, we need to break out the slow-path
(registering the type) into a secondary function that is not a candidate
for inlining.
This ensures that the common case (type registered, return the GType id)
is the hot path and handled in the prologue of the generated assembly even
when -fstack-protector-strong is enabled.
https://bugzilla.gnome.org/show_bug.cgi?id=795180
The existing implementation was completely incorrect (despite the fix in
commit 566e64a66) — it always compared GVariants by pointer, rather than
by value.
Reimplement it to compare them by value where possible, depending on
their type. The core of this implementation is g_variant_compare(). See
the documentation and tests for further details of the new sort order.
This adds documentation and tests.
Signed-off-by: Philip Withnall <withnall@endlessm.com>
https://bugzilla.gnome.org/show_bug.cgi?id=795735
This was causing g_param_value_defaults to return 1
for GVariant values even when the value is clearly
different from the default.
This was showing up as gtk-builder-tool stripping
non-default values for GtkActionable::action-target
from ui files.
It was previously a copy–paste of G_VALUE_COLLECT, which is wrong,
because it’s the inverse function of G_VALUE_COLLECT. Document that, so
the whole thing is a little less confusing (but by no means perfect).
Signed-off-by: Philip Withnall <withnall@endlessm.com>
Reviewed-by: nobody
All those logging functions already add a newline to any message they
print, so there’s no need to add a trailing newline in the message
passed to them.
Signed-off-by: Philip Withnall <withnall@endlessm.com>
Reviewed-by: nobody
The critical omission from the GClosure documentation is that you need
to call g_closure_set_marshal() when implementing a custom GClosure.
Signed-off-by: Philip Withnall <withnall@endlessm.com>
Reviewed-by: nobody
glib-genmarshal shows redundant "time" warning message against combination with --header and --body option.
Before:
WARNING: Using --header and --body at the same time time is deprecated; use --body --prototypes instead
After:
WARNING: Using --header and --body at the same time is deprecated; use --body --prototypes instead
https://bugzilla.gnome.org/show_bug.cgi?id=795429
Signed-off-by: Kentaro Hayashi <hayashi@clear-code.com>
On gcc 4.7, we got the following error:
i686-nptl-linux-gnu-gcc --version
> i686-nptl-linux-gnu-gcc (crosstool-NG 1.20.0) 4.7.4
> $ echo '#include <glib-object.h>' | i686-nptl-linux-gnu-gcc -x c -I
staging/usr/include/glib-2.0 -I staging/usr/lib/glib-2.0/include -Wall
-Werror -c - -o /tmp/foo.o
> In file included from
staging/usr/include/glib-2.0/gobject/gbinding.h:29:0,
> from staging/usr/include/glib-2.0/glib-object.h:23,
> from <stdin>:1:
> staging/usr/include/glib-2.0/gobject/gobject.h: In function
'g_set_object':
> staging/usr/include/glib-2.0/gobject/gobject.h:725:5: error: value
computed is not used [-Werror=unused-value]
> cc1: all warnings being treated as errors
This error has been added by commit 3fae39a5d7
So enable the new g_set_object definition only if gcc >= 4.8
Fixes:
- http://autobuild.buildroot.net/results/b29a2f868438a2210873ea72f491db63175848be
Signed-off-by: Fabrice Fontaine <fontaine.fabrice@gmail.com>
https://bugzilla.gnome.org/show_bug.cgi?id=795138
There is no transfer annotation that can express transfer semantics of
g_object_new_with_properties in general. When GInitiallyUnowned object
is constructed the introspection data will be incorrect.
Mark it with skip annotation.
https://bugzilla.gnome.org/show_bug.cgi?id=795025
Previously we were only detecting typedef\*senum\s*\{, which does not
handle the case where there is an entifier for the enum itself but
not the typedef. glib-mkenums would then attempt to read the next line
looking for a matching {, but in vain.
https://bugzilla.gnome.org/show_bug.cgi?id=794506
It’s defined in gutils.h, but various users of GLib might not have
access to that.
Signed-off-by: Philip Withnall <withnall@endlessm.com>
Reviewed-by: nobody
https://bugzilla.gnome.org/show_bug.cgi?id=793272
The *_init() functions have prototypes incompatible with *InitFunc types they
are being cast to. This upsets GCC 8's -Wcast-function-type that's enabled by
default with -Wextra.
Let's not have the public header files emit a warning and neutralize it by
doing a void(*)(void) cast first.
https://bugzilla.gnome.org/show_bug.cgi?id=793272
It's not possible to subclass GValue, and by always explicitly casting
here it is easy to write broken code (e.g. passing a GValue**) without
the compiler warning about that.
By not casting, the compiler will error out if anything but a GValue* is
passed here.
https://bugzilla.gnome.org/show_bug.cgi?id=793186
gtk-doc doesn’t support them any more since it was ported to Markdown,
so they end up appearing in the generated documentation, which isn’t
great.
Mostly, they were used to split up things invisibly, which we can do in
other ways.
Signed-off-by: Philip Withnall <withnall@endlessm.com>
Reviewed-by: nobody
Properly define GLIB/GOBJECT_STATIC_COMPILATION when static build is enabled.
Use library() instead of shared_library() to allow selecting static builds.
https://bugzilla.gnome.org/show_bug.cgi?id=784995
This makes easier to write a module that can be both dynamic and static.
It will allow to statically build modules from glib-networking, for
example.
A module can rename its g_io_module_load() function to
g_io_<modulename>_load(), and then an application which links statically
against that module can call g_io_<modulename>_load(NULL) to register
types and extension points from the module. If a module is loaded
dynamically, its load() function will continue to be called with a
non-NULL GIOModule instance.
https://bugzilla.gnome.org/show_bug.cgi?id=684282
This is for destroying resources needed by transformations. But the user
may not need any such resources. Make it obvious that, instead of having
to point to a no-op function, @notify is checked and not called if NULL.
https://bugzilla.gnome.org/show_bug.cgi?id=792098
Weak-pointers are currently lacking g_set_object() & g_clear_object()
helpers equivalent. New functions (and macros, both are provided) are
convenient in many case, especially for the property's notify-on-set
pattern:
if (g_set_weak_pointer (...))
g_object_notify (...)
Inspired by Christian Hergert's original implementation for
gnome-builder.
https://bugzilla.gnome.org/show_bug.cgi?id=749527
Conceptually, these functions clearly ought to be fine for a const
structure. This avoids _G_TYPE_CVH (the implementation of
G_TYPE_CHECK_VALUE_TYPE, G_VALUE_HOLDS, G_VALUE_HOLDS_BOXED etc.)
needing to cast to a mutable GValue, which causes
G_VALUE_HOLDS (cv, type) to issue warnings under gcc -Wcast-qual if
cv is a const GValue *.
https://bugzilla.gnome.org/show_bug.cgi?id=734479
Some source files aren't valid utf-8 containing for example
iso8859-1 accented characters in author's names.
Replace invalid data with a replacement '?' character and print a
warning to keep things working.
Based on a patch from Christoph Reiter in
https://bugzilla.gnome.org/show_bug.cgi?id=785113#c20
The type propagation breaks the GRefPtr.h class in WebKitGTK, and in
any case existing C++ code calling the C API will need to perform an
explicit cast, as there's no automatic promotion of pointer types to
and from void*.
Tested-by: GNOME Continuous
https://bugzilla.gnome.org/show_bug.cgi?id=790697
When compiling code that includes gobject.h using GCC with the ISO
standard, the `typeof` keyword is disabled, as it's a GCC extension.
The GCC documentation recommends:
> If you are writing a header file that must work when included in
> ISO C programs, write __typeof__ instead of typeof.
Which is precisely what we're going to do.
Signed-off-by: Emmanuele Bassi <ebassi@gnome.org>
Reviewed-by: Philip Withnall <withnall@endlessm.com>
https://bugzilla.gnome.org/show_bug.cgi?id=790697
Currently, g_object_ref() and g_object_ref_sink() return a
gpointer which can mask issues when assigning to fields or
returning from a function.
To help catch these type of programming errors, we can propagate
the type of the parameter through the function call on GCC
using the typeof() C language extension.
This will cause offending code to have a warning, but will
continue to be source and binary compatible.
This is only enabled when GLIB_VERSION_MAX_ALLOWED is 2.56 or greater.
https://bugzilla.gnome.org/show_bug.cgi?id=790697
Putting a <!-- --> in plural<!-- -->s was an old hack used to fix
linking the symbol with gtk-doc when gtk-doc didn’t know about plural
forms. gtk-doc does now know about plural forms, so the hack can be
removed.
Signed-off-by: Philip Withnall <withnall@endlessm.com>
It's theoretically possible that we could have a case where this would
actually return NULL, but it's difficult to imagine a valid program that
would contain such a case.
Add an explicit assert here to quiet up static analysis.
See the bug for more discussion.
Coverity CID: 1159477
https://bugzilla.gnome.org/show_bug.cgi?id=730296
