We are pleased to announce a new release of the ADL Workbench (‘AWB’) based on the latest ADL / AOM 1.5 draft specifications. This release includes Reference Model checking and visualisation, i.e. validation of archetypes against a machine representation of the openEHR reference model, and numerous improvements in validation, and specialisation-related functionality. It is available for Windows, Mac (Leopard/Intel) and Linux.
See the help and download page.
NEW - Reference Model (RM) Checking
Archetypes are now validated against the openEHR reference model, enabling the AWB to tell if they will actually work in a real openEHR system, not just if they are internally coherent. In this release, the reference model is represented as a human- and machine-readable dADL file obeying a simple object meta-model. The latter is programmed in Eiffel classes, and could easily be ported to other languages. For a discussion on why we have not used XMI or some other form of more UML 2.x based representation, see here. We expect the reference model representation to evolve, and the representation used here should not be taken as normative. Practically speaking, the use of a computable RM has enabled visualisation of RM attributes within the workbench in both differential and flat modes, which can be turned on and off in the tool as required.
Reference Model checking is turned on in the Tools > Options menu.
IMPROVED - Validation Checking
All validity statements defined in the AOM 1.5 draft are now checked, and reported with the validity code, e.g. VCORM, etc.
NEW - ADL save mode
We have added an initial capability to save out flattened ADL archetypes, from the Test page of the tool (right-hand side tab). The ADL version of the output format can be controlled via a setting in the .cfg file, called adl_version_for_flat_output. This could be set as follows:
adl_version_for_flat_output=1.4
If this is not set, the default output version will be 1.5. The main differences setting of 1.4 enables is removal of () around C_DV_QUANTITY and other dADL sections, and also the ‘generated’ keyword is not output. This would allow such archetypes to be processed by non-ADL 1.5 compliant tools.
WARNING: this option should preferably be used in a Subversion work area or similar situation where previous versions can be reverted to if required.
Status & Future
The AWB validation is about 95% synchronised with that of the openEHR Java ADL parser, which is used the Clinical Knowledge Manager, and should soon be 100% compatible. It is also used as the basis for testing all formal elements of the above-mentioned ADL/AOM 1.5 specifications, and will soon do the same for templates defined by the TOM specification.
The a-path sub-language (itself based on W3C Xpath language) developed by Zilics is likely to be the basis of the next version of the assertion language used in archetypes, and the AWB will start supporting this in some form soon.
Support
Currently we still have not moved the ADL Jira project from the Ocean Jira server to the openEHR.org Jira server (this requires upgrades on both Jira installations and then a somewhat tricky dump-and-import operation). However, we will create a new Jira project on openEHR imminently for the purposes of problem reporting.
Acknowledgements
Thanks to…
-
Peter Gummer (Ocean Informatics) for creating the Linux and Mac builds of the tool, as well as finding and fixing numerous bugs;
-
Rong Chen (Cambio Healthcare Systems) for ongoing support of the Java ADL tools, which have often enabled us to find bugs or anomalies in the workbench;
-
Sebastian Garde (Ocean Informatics) who has used the Java tooling within the Clinical Knowledge Manager to find many problems which have further helped us with the workbench;
-
clinical modellers who have given useful feedback.
