ADL 1.5

[jure]

Any plans to support ADL 1.5 in this gem? http://www.openehr.org/downloads/ADLworkbench/learning_about

If not, I'm happy to do a pull request that adds 1.5 support.

[skoba]

I am writing on, but always welcome to merge your pull request.

[skoba]

This is my working repository on ADL15parser. I am testing parslet parser library.

https://github.com/skoba/adl15parser-ruby

[jure]

Thanks, I'll take a look!

[skoba]

I am much sorry to be inactive.

[jure]

No worries, Shinji, you're doing great work with OpenEHR and Ruby!

I looked at the adl15parser-ruby repository and it looks like it's just a skeleton at this point. Any reason why, other than lack of time, you stopped developing it?

There are quite a few changes coming from 1.4: http://www.openehr.org/wiki/display/ADL/Evolution+from+ADL+1.4 but there is a reference compiler implementation here: http://www.openehr.org/wiki/display/ADL/ADL+1.5+parser+resources

There is an effort to write an EBNF description of 1.5 here: https://github.com/aanastasiou/adl_ebnf

What would you say the hardest part for supporting 1.5 will be?

I'm sure you know all of this already, but it might be a good idea to have these references here for posterity.

[skoba]

Two reasons.

1. OPT parser would be emerging. Actually, templates are more available than ADL 1.5 archetypes
2. Flatten form of archetype or ontology. I cannot figure out the flatten form of ontology of ADL/AOM 1.5. I think this is the hardest point.

[skoba]

My ADL 1.4 parser has been written by Treetop parser generator, but it became huge and ugly to read. That 's the reason I am trying to reimplement ADL parser by parslet.

[jure]

Oh, OPT is the template parser. But 1.5 archetypes will be able to be templates as well? Won't that make the OPT parser unnecessary?

It looks like @aanastasiou is trying to come up with an ANTLR definition of ADL, which would enable him to generate a code artifact from the definition. What are your thoughts on that approach?

I think @wolandscat is also very interested in various implementations of ADL in the wild, so I'm pinging him too.

Is there any ADL 1.5 parser out there at all? In any language?

[skoba]

I know only the Eiffel implementation for ADL 1.5 parser. https://github.com/openEHR/adl-tools/tree/master/components/adl_compiler/src/syntax/adl/parser

I think implementation diversity would be a powerful proof of syntax validity, too.

[aanastasiou]

Hello Jure

You probably need to have a closer look at the openEHR specifications (here: http://www.openehr.org/programs/specification/releases/1.0.2) or at least http://www.openehr.org/releases/1.0.2/architecture/overview.pdf cover-to-cover and from there go through the "Archetype Model" documents selectively and gradually. You can also have a look at http://www.openehr.org/wiki/display/ADL/Evolution+from+ADL+1.4 which talks extensively about "specialisation" which is a key concept to understand templates.

Archetypes become usable in templates. Your "top-level" object is the Template, not the Archetype. A template can reference many different archetypes and provides additional constraints to them FOR A SPECIFIC USE.

Is there any ADL 1.5 parser out there at all? In any language?

I second Singhi's response (But, don't forget, we have the specification, therefore, we essentially know the language)

Key question in all of this: What are you trying to do?

[skoba]

I just implemented ADL1.5 parser by parslet, but it needs to transform to generate archetype instances. https://github.com/skoba/adl15parser-ruby