| Commit message (Collapse) | Author | Age | Files | Lines |
... | |
| |
| |
| |
| | |
@modulesdir@
|
|/ |
|
|
|
|
|
|
|
|
| |
TrackerFileNotifier is finished
If the last files in the loop are iterated without success, the last priv->current_index_root
would remain at the time of emitting the ::finished signal, which might confuse
tracker_file_notifier_is_active() afterwards.
|
|
|
|
|
| |
The extractor output has changed. Similar fix can be seen in
commit aabc8a8e07e90b8fae0172824185e50b6af68228.
|
|
|
|
|
|
|
|
|
|
|
|
| |
The tracker_miner_fs_wait_for_idle() call is used to detect miner operation
has completed. It returns when miner's status changes to "Idle" (or on
timeout). Unfortunately at the time miner goes idle it is not guaranteed the
data are already in store - and the related test fails.
A better way is to listen to GraphUpdated signal sent by store and wait until
the desired resource is announced beng added or removed.
Depends on tracker-tests-allow-reuse-graph-updated-signal-handling.patch.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
In many test cases the tracker_miner_fs_wait_for_idle() call is used to detect
miner operation has completed. It returns when miner's status changes to
"Idle" (or on timeout). Unfortunately at the time miner goes idle it is not
guaranteed the data are already in store - and the related test fails.
The test case 301-miner-resource-removal.py does it a better way - it listens
to GraphUpdated signal sent by store and waits until the desired resource is
announced being added or removed.
There is comment inside 301-miner-resource-removal.py: "FIXME: put this stuff
in StoreHelper". This patch is to follow that comment.
Needed by tracker-tests-310-fts-indexing-use-graph-updated-signal.patch.
|
|
|
|
|
|
|
| |
The tests.xml is autogenerated by create-tests-xml.py. It scans all *.py
scripts for test class definitions and generate one test case for each class
found. In that test case it invokes the script with the class name passed as
an argument. 400-extractor.py expect different kind of argument.
|
|
|
|
|
|
|
| |
The process tracker-miner-fs causes the process tracker-store be started
automatically via the D-Bus service autostart mechanism. As the test case
needs to start and control the processes itself, it is necessary to start them
in order of their dependencies, so the D-Bus autostart does not happen.
|
|
|
|
|
|
|
|
|
| |
sometimes
Some test cases need more than the default 90 seconds to complete. As the
tests.xml is autogenerated, it is not possible to increase the timeout
selectively without bigger changes to the create-tests-xml.py script. Thus
increasing globally.
|
|
|
|
|
|
|
| |
The command `tracker-control -t` simply kill(2) all tracker processes and
exits. It does not wait/check the processes to terminate. In some test cases
it happens that the test case tries to launch its own instance of some tracker
process, but it fails because the old process is still there.
|
| |
|
| |
|
| |
|
|
|
|
|
| |
If the service initialization fails, a NULL object will be returned,
so check for that before attempting to communicate with the thumbnailer.
|
|
|
|
|
|
|
|
|
|
|
| |
Before commit 73c0decd, src/libtracker-extract/Makefile.am was receiving those
indirectly from $(topsrcdir)/Makefile.decls. After that file was no longer included
those remained empty, causing wrong file lookups in the built
tracker-miner-fs and tracker-extract.
So the declarations have been moved to a separate $(topsrcdir)/Makefile.extractdirs,
so they're defined once and available to both src/libtracker-extract/Makefile.am and
src/tracker-extract/Makefile.am.
|
|
|
|
|
|
|
| |
Don't install libtracker-extract system-wide and don't generate any
introspection data.
This is a follow-up commit to 60fdc9b900ae1c66a65721c8314e8cc3358ca461
|
|
|
|
|
|
| |
So we don't try to free invalid pointers again on finalize()
https://bugzilla.gnome.org/show_bug.cgi?id=724984
|
| |
|
| |
|
| |
|
| |
|
| |
|
|
|
|
| |
Missing TRACKER_ prefix in enum value...
|
| |
|
|
|
|
|
|
|
|
|
| |
Specifically:
- Fix offset error when highlighting parts of the tree when -s is used
- Use GNode APIs g_node_traverse() to print and find nodes, reducing code size
- Remove unnecessary memory allocation for --get-longhand
- Fix memory leak of longhand to shorthand conversion in --tree
- Avoid memory allocations when generating filter hashtable
|
| |
|
|
|
|
|
|
|
|
| |
Specifically
--tree $CLASS (to print the tree for $CLASS, or NULL for all ontology)
--get-shorthand $CLASS
--get-longhand $CLASS
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
We imported Makefile.decl from glib when it was using Subversion, a long time
ago. The script had a few issues (like running all unit tests twice on make
distcheck) and needed to be updated.
I've stolen a latest version from GLib's glib.mk and also their m4 macros
which are required to make this all work too.
This gives us the advantage of defining installable and uninstallable test
cases as well as test data, scripts and other useful stuff.
The debug output also looks quite good.
And distcheck passes :)
|
| |
|
|
|
|
|
| |
Messages were marked as translatable there, so it needs to be in
POTFILES.in
|
|
|
|
|
| |
A few headers were missing in SOURCES, and noinst_HEADERS was mistakenly
empty, which broke build if builddir!=srcdir
|
| |
|
|
|
|
|
| |
Fixes the warnings about GError and other symbols which are unknown when
generating the documentation for libtracker-{sparql|miner|control}
|
|\ |
|
| | |
|
| | |
|
| |
| |
| |
| | |
Will be useful for long term ABI compat.
|
| | |
|
| | |
|
| |
| |
| |
| |
| |
| |
| |
| |
| | |
TrackerMinerWeb vmethods are very focused on the management of
credentials and capabilities, this is something that's been taken
over by gnome-online-accounts since this API was added, and the
lack of TrackerMinerWeb implementations in the wild seems to
hint it's not been as useful as when it was initially devised,
so farewell.
|
| |
| |
| |
| | |
This is more suitable than implementing TrackerMiner directly
|
| |
| |
| |
| |
| |
| |
| |
| |
| | |
This TrackerMiner is a simpler replacement for TrackerMinerWeb that
doesn't get into credentials handling. It handles network state,
emitting ::connected or ::disconnected on the way, and ensuring the
miner is paused/resumed as necessary when suitable networks come and
go, implementations of this miner can control this behavior through
the return value in the ::connected signal.
|
| |
| |
| |
| |
| |
| | |
TrackerMinerManager has been moved from libtracker-miner, so it remains a library
to implement miners. libtracker-control will be of interest to anything that
attempts to track or control the state of running/available miners.
|
| |
| |
| |
| |
| | |
TrackerMinerManager isn't used here after all, so remove all code
using it.
|
| |
| |
| |
| |
| |
| | |
This object can remain entirely private, as it is wrapped in quite more
convenient ways by TrackerMinerFS (ie. tracking of additions/removals,
file monitoring...), so turn it fully private.
|
| |
| |
| |
| |
| |
| |
| | |
This is used in a variety of places (tracker-writeback, TrackerMinerFiles,
and TrackerDecorator), but it doesn't make sense to have that exposed in
libtracker-miner, so move to libtracker-common, where it can be used in a
private manner.
|
| |
| |
| |
| | |
This object just needs it to iterate over .desktop files, so use plain GIO there.
|
| |
| |
| |
| |
| | |
This can be easily replaced by a helper function that iterates
over a directory.
|
| |
| |
| |
| |
| |
| |
| |
| | |
These calls to manipulate the media art queue are already
performed inside TrackerMinerFS, and there are no foreseeable
situations where a TrackerMinerFS implementation will need
to call these directly, so just move the code and header
to private land.
|