You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
execute_query_on_graph now skips nodes that are not part of the output
(optional negated nodes) and makes sure the resulting iterator only produces
unique results.
Queries with @ could have extremly slow execution plans when the query
planner introduces an inverted @ operator and miscalculated the cost
compared to the non-inverted version.
Frequency queries now execute the (additional) timeout check after a certain
number of matches are processed, not if a specific tuple value has reached a
treshold.
💥 Removed usage of the smartstring crate. This changes the interface of
the AnnoKey, Annotation, and Component/AnnotationComponent structs and
is a breaking change, but the new interface is actually easier to call
because the conversion to smart strings is not necessary any longer. The
change also increases performance for several queries in the test set by up to
30%.
Added
Added the support for Facet to some of the core types
(like Annotation, Component, etc.) to make it possible to use Facet in the Annatto
Avoid loading the node annotation storage when listing the components for a
corpus in the CorpusStorage. Before this change, querying for components via
the webservice could block the corpus cache.
Better estimation for queries with regular expressions without prefix.
Compile releases of the C-library on Ubuntu 22.04 instead of 20.04, which means the minimal
GLIBC version is 2.35. This is necessary, since GitHub actions deprecated this
Ubuntu version.
Added
New optional file option for the [logging] section in the webservice
configuration. Can be used to additionally output all log messages to the given
file.
Add number of root nodes to graph storage statistics. This changes the way
most of the graph storages store their statistics. You can use old imported data
files, but to make use of the new information you queries, you have to reimport your corpora.
Graph:ensure_loaded_parallel returns the actually loaded components that did
exist.
Fixed
Less frequent corpus cache status updates in log. Before, every corpus access
could trigger an entry into the log which is not desired under heavy load.
Improve query execution planning by assuming all annotations can be matched in
regular expressions without a prefix.
Fix broken imports for existing corpora when they never have been added to the
cache or have been evicted from it. (by https://github.com/matthias-stemmler)
Deprecated
Graph::load_from is replaced with the open and import methods.
Added
Allow to add updates to the annotation Graph without re-calculating the
statistics with apply_update_keep_statistics. This is useful for scenarios
were we assume the changes don't change the graph that much and we want to
apply the updates as fast as possible.
Open an Graph from an external location with open or import the changes
into the current graph.
