| Commit message (Collapse) | Author | Age |
| |
|
| |
|
| |
|
|
|
|
|
| |
* Use US English spelling consistently
* Avoid Eclipse Collections types on API
|
| |
|
| |
|
|
|
|
|
|
|
|
|
|
| |
Merge the candidate may/must queries with the partial may/must queries so that
reasoning in the candidate model uses "Gentzen-style" assumptions about the
candidate values of partal symbols that would be computed as unknown.
This should not lead to performance degradation, because the corresponding joins
were already being computed in the stopping criterion. In many cases, the extra
clauses are immediately optimized away by the query optimizer anyways.
|
|
|
|
|
| |
References marked as partial are not concretized during model generation. The
should be managed by the user manually using propagation rules instead.
|
| |
|
|
|
|
|
| |
Switch to partial interpretation based neighborhood calculation when multiple
models are request to avoid returning isomorphic models.
|
| |
|
| |
|
|
|
|
| |
Require all abstract domain elements to implement the AbstractValue interface.
|
| |
|
|
|
|
|
|
|
| |
Efficiently store cross references even if default ref(*, *): false.
Uses propagation rules instead of view queries to reason about type constraints
and multiplicity.
|
|
|
|
| |
We were spuriously counting lower multiplicity errors for non-existent objects.
|
| |
|
|
|
|
|
|
|
|
|
|
| |
Now unary predicates may be used as types for cross-references.
This might create circular references between predicate, but no differently than
when the user manually creates predicates with circular references.
The reference types of containment and container references still have to be
classes, because we use the class hierarchy to encode contained types.
|
|
|
|
|
|
|
|
| |
Rename VIATRA subprojects to Refinery Interpreter to avoid interfering with
Eclipse Foundation trademarks.
Uses refering to a specific (historical) version of VIATRA were kept to avoid
ambiguity.
|
| |
|
|
|
|
|
| |
If an object has a lower bound of 1, but it is still a multi-object, model
generation can't be finished yet.
|
| |
|
| |
|
| |
|
| |
|
| |
|
| |
|
| |
|
|
|
|
|
| |
Use a simplified name invalidContainer, since this pattern may match even if
there is a sufficient number of ERROR containers.
|
|
|
|
|
|
| |
Providing an efficient refinement calculus would be too difficult at this point,
and contains is much more useful as a constraint in predicates that as an
assertions anyway.
|
|
|
|
| |
Make sure to merge all clusters reachable from source and target.
|
|
|
|
|
| |
Also makes model seeds cancellable to reduce server load during semantic
analysis.
|
| |
|
| |
|
|
|
|
| |
Avoid creating interpretations not needed for the graph analysis task.
|
| |
|
|
|
|
|
| |
Improper equality implementation of representative election literals merged weak
and strong connected components during query canonicalization.
|
| |
|
| |
|
|
|
|
|
| |
Reduce memory use of initial partial models by consolidating value
objects.
|
| |
|
| |
|
| |
|
|
|
|
|
|
|
|
|
| |
Adds translator for EQUALS and EXISTS symbols based on the multi-object
formalism. Only diagonal equality links are supported (e.g., distinct nodes may
not be EQUALS with each other).
Also introduces initial model seeds to separate partial interpreter construction
and graph initialization better.
|
| |
|
|
|
|
|
| |
* Use method object pattern to simplify code.
* Optimize existential quantification in transitive queries.
|
|
|
|
| |
Support for count and aggregation is still missing.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Enable deterministic state-space exploration by ordering activations in
lexicographic order.
This preliminary implementation adds oredering as a wrapper for ResultSet
instances, but more sophisticated support could be built directly into query
engine adapters if a query engine supports deterministic output by default.
* Implements Comparable for tuples with loops unrolled for small tuples by hand.
* Cleans up the contents of the (root of the) tools.refinery.query package.
* Adds ResultSetListener to notify clients about ResultSet changes.
* Adds OrderStatisticTree data structure for determinisitc ordering of keys.
|
|
|
|
|
| |
The filer can't return true for the default value, because tuples with default
value cannot be enumerated.
|
|
|
|
| |
We will need to rework rule execution according to recent Dnf changes.
|