For those interested in ADL/AOM 1.5, I made a post in my blog giving a progress update. It’s getting simpler, more powerful, and more generic.
- thomas beale
I wonder, will it still be compatible with the ADL 1.4 specifications? Bert
The specification is technically no longer backward compatible, primarily because the special C_DV_ORDINAL and C_DV_QUANTITY syntaxes are no longer supported - in fact the whole openEHR Archetype profile is obsolete in ADL 1.5. However, tools can easily convert from 1.4 to 1.5, and I have made the ADL Workbench work like this. The programming patterns for supporting these ADL 1.4 special types are easy enough and can be copied from my code into Java or any other parser. To help with retro-fitting this kind of functionality, I’ll document the relevant parts of the parser on the wiki page, and am happy to provide any guidance on it. - thomas
It is a good thing, I always hated these syntaxes.
But of course, in my case, software must be able to use/process old archetypes.
But I guess this must be solved then by distinguishing the archetypes by version, and take precautions in further processing software.
A conversion-routine from the profile-class-instances to the new AOM class-instances is maybe able to do the trick, do you think that will be feasible?
Bert
The specification is technically no longer backward compatible, primarily because the special C_DV_ORDINAL and C_DV_QUANTITY syntaxes are no longer supported - in fact the whole openEHR Archetype profile is obsolete in ADL 1.5.
It is a good thing, I always hated these syntaxes.
But of course, in my case, software must be able to use/process old archetypes.
But I guess this must be solved then by distinguishing the archetypes by version, and take precautions in further processing software.
yes - ADL version is always indicated at the top of an archetype, even and ADL 1.4 archetype, so this can be used reliably.
A conversion-routine from the profile-class-instances to the new AOM class-instances is maybe able to do the trick, do you think that will be feasible?
certainly. In fact, at some point soon, I'll create a command line converter to batch convert ADL 1.4 archetypes to ADL 1.5 archetypes for the 3 main platforms and publish that.
- thomas
Update on progress in the last few weeks:
I have implemented a reasonable form of the value-sets in terminology idea. All is explained here on the updated wiki page. Originally I had notions of making some ‘powerful’ triple-based design, but in the end, a more pragmatic solution more or less like Daniel Karlsson proposed made more sense. The wiki page has examples, so you can see easily what’s going on.
The terse summary of all changes in recent times is on this page.
I am working with Harold Solbrig @ Mayo on refining the last details of the terminology binding / value set AOM model semantics. A completely up to date set of specification documents will be available very soon. The AOM UML shown here is not quite up-to-date, but is very close, in case you are interested.
The ADL Workbench will also be available in a new limited release imminently. It has taken some time for me to implement the conversions described on the above page, so that archetypes you may already have will be completely or nearly completely converted automatically for you.
A lot of these changes have come from working with CIMI, and I would say we have made more progress on clinical modelling in the last few months than in the last 3 years. Nothing like some cross-fertilisation…
- thomas