This is an incremental update to the 2026-07 release. It corresponds to Eclipse plugin version 0.62.0.
Language Features
None.
Model Libraries
None.
Backward Incompatibilities
None.
Issue Resolutions
None.
Jupyter
None.
Visualization (PlantUML)
None.
Technical Updates
-
Caching. The implementation uses caching of various computed values in order to improve performance. However, this means that, when a derived value visible externally is cached, it's value will not change when inputs to the derivation change, unless the cache is first cleared. In particular, the
featureMembershipsof an element are cached. Previously, this meant that, ifgetFeatureMembershipwas called, and then a newFeatureMembershipwas added, this was not reflected in additional the result of subsequent calls togetFeatureMembership, unlessclearCacheswas explicitly called first.This has now been changed so that
getFeatureMembershipautomatically clears caches of the element being accessed before computing thefeatureMembershipsof that element. In order to retain the performance benefit of caching, internal code has been revised to use the utility methodTypeUtil::getFeatureMembershipsOfinstead ofType::getFeatureMembership, which preserves the previous caching behavior. In addition, similar changes have been made on the getters for several other derived properties computed based onfeatureMembership– for details, see the PR. There are also other "families" of getters for other cached derived property values, but these have not yet been addressed.[PR #773]
-
Parse post-processing. After the completion of Xtext parsing of a model file, the implementation carries out several post-processing computations to correct for limitations in the KerML and SysML grammars, as specific using the Xtext grammar language. Conceptually, this post-processing can be consider to be be part of the parsing phase when processing model text. However, previously, the post-processing code was added to the
ElementAdapters, which mixed this Xtext-specific code with the Xtext-independent semantic transformation code which is the main purpose of these adapters. In this release, the post-processing code has been removed from the adapters in the packageorg.omg.sysml.logic.adapterand moved to two new packages,org.omg.kerml.xtext.postprocessingandorg.omg.sysml.xtext.postprocessing.[PR #793]
Bug Fixes
- Owned disjoining. Fixes a bug that caused
Type::getOwnedDisjoiningto always return the empty list.
[PR #791] - Implicit subsetting. Corrects the removal of duplicate implicit subsetting relationships that target feature chains.
[PR #792] - Port usage. Corrects
PortUsageAdapterto extendOccurrenceAdapterrather thanUsageAdapter.
[PR #795] - Cross-subsetting. Fixes a bug which caused the
KerMLValidatorto throw an exception when checking invalid cross-subsettings.
[PR #796] - Composite usages. Revises the handling of
isCompositeandisReferencefor usages, so thatisComposite = falseis preserved on an XMI roundtrip.
[PR #797] - Feature types. Revises
FeatureAdapter::getTypesso it doesn't throwNullPointerExceptionwhen deriving the type of aFeaturewhosetypingFeatures()result contains anullentry.
[PR #798]