| Commit message (Collapse) | Author | Age | Files | Lines |
|
|
|
|
|
|
|
|
|
|
| |
This may happen during query evaluation, without the query explicitly
specifying any NULL value. If we error out, query execution will
definitely stop, and the error may propagate further up.
Just tiptoe over those values, and let query evaluation continue
further, the right results will be eventually returned.
Fixes: https://gitlab.gnome.org/GNOME/tracker/-/issues/235
|
|
|
|
|
|
|
|
|
|
|
| |
This breaks fundamental assumptions (e.g. that getting a next resource ID is
not failable), breaks even further if those resources don't get the specified
ID.
It is preferrable to hard error altogether, and let users report the
backtrace/error.
Related: https://gitlab.gnome.org/GNOME/tracker-miners/-/issues/150
|
|
|
|
|
| |
Instead of warning in place, let these errors propagate up, so they're
handled centrally.
|
|\
| |
| |
| |
| | |
website/faq: Add note about RAM usage
See merge request GNOME/tracker!334
|
|/ |
|
|\
| |
| |
| |
| | |
website: FAQ updates
See merge request GNOME/tracker!329
|
| |
| |
| |
| | |
See https://gitlab.gnome.org/GNOME/tracker/-/merge_requests/329#note_933276
|
| |
| |
| |
| |
| | |
Use 'Tracker Miner FS' instead of 'Tracker' in more places,
and `tracker3` instead of `tracker`.
|
|/ |
|
|\
| |
| |
| |
| | |
Tracker3 updates for README and website
See merge request GNOME/tracker!331
|
| | |
|
| | |
|
|\ \
| | |
| | |
| | |
| | | |
Improve behavior on parallel queries
See merge request GNOME/tracker!333
|
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | | |
We typically do that, but once we've found that the first interface
in the pool is already busy with other statements, we assume they're
all busy and jump into creating a new interface. Even worse, if that
interface is kept busy indefinitely, every new query will opt for
creating a new interface, quickly filling the allotment.
There might be other interfaces in the pool that are already
free, so check all of them before trying to create a new interface.
This makes us more conservative at creating new interfaces on load
periods, only doing so if all interfaces are actually busy.
|
|/ /
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
If we happen to get a high enough amount of updates, batched changes
accumulate, all querying for the additional information. This
accumulation of the same query leads at best to compiling new
statements (as the cached one is already in use). At worst, it leads
to the creation of a new TrackerDBInterface, with the initialization
cost involved.
Serializing this work, we're more likely to reuse a TrackerDBInterface
and a pre-compiled statement from previous runs.
|
| |
| |
| |
| | |
It broke after c32f172f1b48f6cfe9845ca233f643bde81abd0c.
|
|\ \
| | |
| | |
| | |
| | | |
libtracker-data: Change GPtrArray into GenericArray
See merge request GNOME/tracker!332
|
|/ /
| |
| |
| | |
This is the vala-preferred way of handling GPtrArray
|
|\ \
| |/
|/|
| |
| | |
docs: Add information related to Flatpak
See merge request GNOME/tracker!330
|
| |
| |
| |
| |
| |
| |
| |
| | |
Also some general cleanups, in particular removing domain-ontology
feature which now belongs in tracker-miners.git, and can be
documented there with an example app.
See: https://gitlab.gnome.org/GNOME/tracker/-/issues/236
|
|\ \
| |/
|/|
| |
| | |
website: Fix broken link in preview API docs
See merge request GNOME/tracker!328
|
| |
| |
| |
| |
| |
| | |
This fixes a broken link that was in the header of each page of the
preview API documentation, pointing to
https://gnome.pages.gitlab.gnome.org/tracker/docs/ which gives a 404.
|
| |
| |
| |
| | |
Next stable release will be 3.1.0.
|
|\ \
| | |
| | |
| | |
| | |
| | |
| | | |
libtracker-data: Process Update rule iteratively
Closes tracker-miners#91
See merge request GNOME/tracker!327
|
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | | |
The Update rule is defined upon itself, we interpret this a bit too
literally, and do the same thing when interpreting the parse tree. This
makes the maximum stack size an indirect factor that limits how big a
series of updates can possibly be. (e.g. the array at
tracker_sparql_connection_update_array_async)
This is obviously bad, so process the updates iteratively, this will avoid
hitting stack limits by just concatenating legit updates together.
Fixes: https://gitlab.gnome.org/GNOME/tracker-miners/-/issues/91
|
|\ \ \
| | | |
| | | |
| | | |
| | | |
| | | |
| | | | |
libtracker-data: Break out of all loops on transaction errors
Closes tracker-miners#130
See merge request GNOME/tracker!326
|
| |/ /
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | | |
If an error is found when flushing a transaction on a specific, we'd
inadvertently still try to handle operations in other graphs, possibly
reusing the GError location, and leading to invalid reads/writes.
After finding an error, the transaction is going to be rolled back
anyway, so break on the first error found.
Fixes: https://gitlab.gnome.org/GNOME/tracker-miners/-/issues/130
|
|\ \ \
| | | |
| | | |
| | | |
| | | |
| | | |
| | | | |
Fixes to date/time parsing
Closes tracker-miners#146
See merge request GNOME/tracker!324
|
| | | |
| | | |
| | | |
| | | | |
To ensure these work as intended.
|
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | | |
Expand the %F in strftime() to %4Y-%M-%D, otherwise for years < 1000
we end up eating digits, and producing a not quite ISO8601 string.
Fixes: https://gitlab.gnome.org/GNOME/tracker-miners/-/issues/146
|
| | | |
| | | |
| | | |
| | | |
| | | | |
Let this helper SQLITE function forward the datetime conversion, should
there be one.
|
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | | |
When parsing iso8601 strings, we may end up with a legit negative
timestamp. Checking for it being negative makes us fail (with no
error!) for dates before the epoch, check the error instead.
Fixes parsing issues with queries with ancient datetimes like:
select ("0100-12-31T21:00:00-03:00"^^xsd:dateTime as ?date) {}
|
| |/ /
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | | |
When creating a binding from a literal datetime value, we mistakenly
use doubles (as used to be the common thing in Tracker 2.x). We now
store datetimes as either iso8601 strings or int64 timestamps, so
forward these types.
This code is in consistence with the update bits at tracker-data-update.c.
Fixes select queries like:
SELECT ("2020-01-01T01:01:01Z"^^xsd:dateTime AS ?date) { }
|
|\ \ \
| | | |
| | | |
| | | |
| | | |
| | | |
| | | | |
Wip/carlosg/ttl parser improvements
Closes #260
See merge request GNOME/tracker!323
|
| | | |
| | | |
| | | |
| | | |
| | | | |
We don't test directly for LOAD, so squeeze these sneaky cases
(long strings and langtags) in a test where we load a TTL file.
|
| | | |
| | | |
| | | |
| | | |
| | | | |
Instead of forwarding plain strings, forward the langtag info so it
is stored.
|
| | | |
| | | |
| | | |
| | | |
| | | | |
This function takes a string and langtag, and produces a GBytes as
the internals do expect it.
|
| | | |
| | | |
| | | |
| | | | |
Pass this as an argument, unused so far.
|
| | | |
| | | |
| | | |
| | | |
| | | |
| | | | |
Allow LANGTAG after STRING_LITERAL* and STRING_LITERAL_LONG. We don't
propagate the language tag so far yet, just parsing of these tags is
fixed.
|
| |/ /
| | |
| | |
| | |
| | | |
The STRING_LITERAL terminal may mistakenly match STRING_LITERAL_LONG strings.
Invert the order here, so we correctly trimp quotes around.
|
|\ \ \
| | | |
| | | |
| | | |
| | | | |
Updates to 'overview' page on website
See merge request GNOME/tracker!321
|
| | | | |
|
| | | |
| | | |
| | | |
| | | |
| | | | |
This is a bit less easy to read but hopefully makes it clearer that
Tracker SPARQL and Tracker Miner FS are separate things.
|
| | | | |
|
|\ \ \ \
| |_|/ /
|/| | |
| | | |
| | | | |
build: Remove forgotten network_manager option
See merge request GNOME/tracker!325
|
|/ / /
| | |
| | |
| | | |
It is not used since https://gitlab.gnome.org/GNOME/tracker/commit/a88e0f23df07308670020926285549b0ed8a55a2.
|
|\ \ \
| | | |
| | | |
| | | |
| | | | |
portal: Fix initialization order
See merge request GNOME/tracker!322
|
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | | |
The portal is currently first claiming the bus name, then adding
the portal object+interface. This may break things with autostart
as clients are able to send a message to an object path that is not
there yet.
Changing the order means the object path is there when the DBus name
is made known, so clients are able to talk immediately to it.
Fixes the error reported at
https://github.com/flathub/org.gnome.Music/pull/24#issuecomment-702565846
|
|/ / / |
|
| | | |
|