| Commit message (Collapse) | Author | Age | Files | Lines |
... | |
| | | |
| | | |
| | | |
| | | |
| | | | |
Don't call _init_token(). We don't use it underneath this function, and it will
be later filled in with the property path for when we need it in _add_quad().
|
| | | | |
|
| | | | |
|
| | | | |
|
| | | | |
|
| | | | |
|
| | | | |
|
| | | |
| | | |
| | | |
| | | | |
Use g_clear_object on priv->tree to avoid memory leaks.
|
| | | |
| | | |
| | | |
| | | | |
The value returned from a call to g_strjoinv has to be freed using g_free.
|
| | | |
| | | |
| | | |
| | | |
| | | | |
Removing elements from a native vala array can cause memory leaks. Let's use
GLib.Queue both for subject_stack and subject_stack.
|
| | | |
| | | |
| | | |
| | | | |
An array of TrackerGrammarRule has to terminate with a NIL.
|
| | | |
| | | |
| | | |
| | | | |
Make sure db_journal_reader_clear is called in order to free allocated memory.
|
| | | | |
|
| | | |
| | | |
| | | |
| | | |
| | | | |
Use g_value_set_static_string instead of g_value_set_string to avoid memory
corruption.
|
| | | | |
|
| | | |
| | | |
| | | |
| | | | |
Make sure to call g_free on result in tracker-type-utils-test.c
|
| | | |
| | | |
| | | |
| | | | |
Make sure to call g_free on result in tracker-file-utils-test.c
|
| | | | |
|
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | | |
Make sure to use GPOINTER_TO_UINT when using g_hash_table_lookup_extended() to
prevent stack smashing. This will make sure that in the architectures where
sizeof(GFileMonitorEvent) < sizeof(gpointer), g_hash_table_lookup_extended()
will not write more bytes than prev_event_type can hold.
Fixes: https://gitlab.gnome.org/GNOME/tracker/issues/71
|
| | | |
| | | |
| | | |
| | | |
| | | | |
We fallback on libuuid with older glibs, but the #ifdef got the
condition wrong.
|
| | | | |
|
|\ \ \ \
| | | | |
| | | | |
| | | | |
| | | | | |
build: Raise glib dep to 2.46.0
See merge request GNOME/tracker!57
|
| | |_|/
| |/| |
| | | |
| | | |
| | | |
| | | |
| | | | |
It used to be 2.44.0 in configure.ac, and the code does use at least
g_strv_contains, which is new in 2.44.
It now also uses g_async_queue_push_front_unlocked and
g_settings_schema_list_keys, which are new in 2.46.
|
|\ \ \ \
| | | | |
| | | | |
| | | | |
| | | | | |
libtracker-common: improve test_path_evaluate_name environment handling
See merge request GNOME/tracker!59
|
| |/ / /
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | | |
Under some environments $PWD might not point where we expect, so simply
use g_get_current_dir() instead.
g_getenv() is documented to return a pointer that may get overwritten
by subsequent calls to g_getenv, g_setenv of g_unsetenv. As even after
removing the second g_getenv, there's still g_setenv calls, take a copy.
|
|\ \ \ \
| | | | |
| | | | |
| | | | |
| | | | | |
build: Use meson feature option type where appropriate
See merge request GNOME/tracker!61
|
| |/ / / |
|
| | | | |
|
|/ / / |
|
| | | |
|
| | | |
|
| | | |
|
| | | |
|
| | |
| | |
| | |
| | |
| | |
| | |
| | | |
These were added in sqlite 3.21.0. For the places where updating
SQLite is not easy/possible, just define those. Presumably older
SQLite versions will fall back through other operators in the
xBestIndex method, either way we won't get those from the DB engine.
|
| | |
| | |
| | |
| | |
| | |
| | | |
For the lulz, releng all over the place crapped out with the "new"
versioning. This is just a tag, and does not get a tarball in
gnome FTP.
|
|\ \ \ |
|
| | | |
| | | |
| | | |
| | | |
| | | | |
If the query is not cacheable, we were adding string literals unquoted.
Oops.
|
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | | |
There were some special cases that were handled individually:
- select * { <s> ?p ?o } was handled through querying the rdf:types of <s>,
and performing an UNION of all related tables. This doesn't play along
with parameterized variables where the subject might be a variable assigned
later through the TrackerSparqlStatement.
- select * { ?s ?p <o> } used similar tricks (querying <o> this time). It had
all those drawbacks, plus it only worked for properties with rdfs:Resource
range.
- select * { ?s ?p ?o . ?p rdfs:domain <dom> } was handled specially.
- select * { ?s ?p ?o } is unsupported
- select * { <s> ?p <o> } was silently unhandled
All those combinations are transparently handled with the tracker_triples
virtual table now.
|
| | | |
| | | |
| | | |
| | | |
| | | | |
This table may be used in either select or update queries, so make sure it
gets initialized in all paths.
|
|/ / /
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | | |
This eponymous virtual table is able to decompose the full database
in all its composing triples. This may be used to implement queries
with predicate variables in a generic way, and finally support the
kind of queries where we gave up (e.g. "select * { ?s ?p ?o }").
Internally it works by using the TrackerOntologies in order to
split the query into a set of queries for individual
properties/columns, some optimizations happen when specific matches
are given, and the SQLite engine does take care of the ones we don't
optimize, sorting, etc...
This virtual table will also be useful in the future when implementing
CONSTRUCT/DESCRIBE commands.
|
| | | |
|
| | | |
|
| | | |
|
| | | |
|
| | | |
|
| | | |
|
|\ \ \
| | | |
| | | |
| | | | |
See: https://gitlab.gnome.org/GNOME/tracker/merge_requests/53
|
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | | |
Make sure to use GPOINTER_TO_UINT when using g_hash_table_lookup_extended() to
prevent stack smashing. This will make sure that in the architectures where
sizeof(GFileMonitorEvent) < sizeof(gpointer), g_hash_table_lookup_extended()
will not write more bytes than prev_event_type can hold.
Fixes: https://gitlab.gnome.org/GNOME/tracker/issues/71
|
| |/ /
|/| | |
|
| | | |
|