Sometimes ninja decides to compile tests/interactive-panels/main.c,
which requires shell/resources.h, before shell/resources.h is generated.
The fix is to use 'sources' argument of declare_dependency to properly
declare the dependency on generated headers.
A fix for tests/datetime and panels/datetime is also included in the
commit to fix a similar problem.
Closes https://gitlab.gnome.org/GNOME/gnome-control-center/merge_requests/472
This is solely used for filtering by unittest, however, if it is
missing, then unittest will fail. So generate a reasonable name (which
is unfortunately not identical to the real name of the attribute).
The purpose of this interactive test is to be able to
prototype new features for CcPanel without having to
adapt existing panels.
Three test panels were added as part of this initial
patch:
* Dynamic panels: panel exercising the ability to
show and hide itself at runtime.
* Header: panel that embeds a header widget.
* Static Initializer: panel that sets a static init
function to run at startup.
They're broken on my computer, and Python does a really
terrible job at communicating why. So after 8 hours of
debugging, I decided this is just a lost battle to me
and not important enough.
The timezone information is allocated with `tz_info_from_location` but
never deallocated.
LeakSanitizer complains about this memory leak.
```
Direct leak of 13568 byte(s) in 424 object(s) allocated from:
/build/gcc/src/gcc/libsanitizer/asan/asan_malloc_linux.cc:95
#1 0x7f63899bab21 in g_malloc0 (/usr/lib/libglib-2.0.so.0+0x51b21)
../tests/datetime/test-timezone-gfx.c:30
#3 0x7f63899dc76a (/usr/lib/libglib-2.0.so.0+0x7376a)
```
The directory opened with `g_dir_open` is never closed, so LeakSanitizer
complains about a memory leaks.
```
Direct leak of 8 byte(s) in 1 object(s) allocated from:
/build/gcc/src/gcc/libsanitizer/asan/asan_malloc_linux.cc:86
#1 0x7f2012908ac9 in g_malloc (/usr/lib/libglib-2.0.so.0+0x51ac9)
#2 0x7f2012922aac in g_memdup (/usr/lib/libglib-2.0.so.0+0x6baac)
#3 0x7f20128ed2d5 (/usr/lib/libglib-2.0.so.0+0x362d5)
#4 0x7f20128ed338 in g_dir_open (/usr/lib/libglib-2.0.so.0+0x36338)
../tests/datetime/test-endianess.c:29
#6 0x7f201292a76a (/usr/lib/libglib-2.0.so.0+0x7376a)
#7 0x7f201292a69a (/usr/lib/libglib-2.0.so.0+0x7369a)
(/usr/lib/libglib-2.0.so.0+0x73951)
#9 0x7f201292a972 in g_test_run (/usr/lib/libglib-2.0.so.0+0x73972)
#10 0x560ad630cbff in main ../tests/datetime/test-endianess.c:66
```
Python isn't always installed in /usr/bin. If we hard-code the path,
these tests will fail on *BSD because they install python in different
prefixes such as /usr/local and /usr/pkg.
Also wait for the connection-added signal instead of only waiting for the
property notification for the connections. While not neccessary, this is
more correct.
Doing so means we handle the same events as the panel. This probably
makes no difference, but this way it is guaranteed we are not getting
weird inconsistencies during testing.
Also fixes a NMClient memory leak in the process.
This has the side effect of showing UI elements that should not be
visible at startup. Just add the correct gtk_widget_show calls to show
all relevant widgets.
First of all, this is a complete rewrite of the
timezone tests. Everything was revisited, starting
from code style, to concepts, etc.
The problem with the previous timezone test was that
is was relying on listing the /usr/share/zoneinfo
directory, and assuming that those entries would be
always present.
Turns out, some of them are extensions, some of them
are undocumented files, etc. A huge mess. I could've
blacklisted the undesired files and folders, but that
would still be insufficient for other OSes like *BSDs
and Sun.
The final solution was pretty straightforward: only
use the information from zone.tab to run the tests.
This adds tests for the network panel based on the test service found in
NetworkManager. Another possible solution may be to use the one from
dbusmock, however NetworkManager already has readily available code to
write tests in C which makes checking the widget hierarchy easier.
This makes running glib based tests inside a dbusmock environment easier
and more beautiful (i.e. output is supressed unless an error occurs).
This helper has been submitted for inclusion in dbusmock. If it cannot$
live there in some form, then we should try to find a home in the GNOME$
project for it.$
This helper has been submitted for inclusion in dbusmock. If it cannot
live there in some form, then we should try to find a home in the GNOME
project for it.
This adds a bit of new functionality to the test service to set the
status and reason for the status change as well as whether the carrier
is connected.
