Hi Pablo,
for clinical modellers I completely agree. It is mostly technical people
- tool writers who work with the syntax form of things. But at the end
of the day, it is them who build the systems and who must understand
every last subtlety of the semantics of any of the languages we use -
ADL, AQL, ODIN, SNOMED constraint syntax, just as for the mainstream
languages they use, i.e. Java, C#, OWL, and so on.
Determining a clean syntax for any part of a specification is part of
designing what that specification is about (for specifications that have
a syntax aspect). At the moment ADL, ODIN, BMM, and AQL are all
cleartext context-free syntaxes. Yes they tend to be read and written by
tools in operational circumstances, but to ignore the syntax is to
ignore the activities of learning, developing, testing, and debugging.
Imagine trying to teach someone programming with no recourse to a
programming language, only in-memory compiler structures.
Getting language right corresponds to obtaining clarity in a formalism.
Working on the tools is of course a big priority, but a different
exercise and (generally) people - it's a software development exercise.
But they are linked. I'll give you an example. To implement an archetype
flattener properly, as in the ADL Workbench, you need to know an
algorithm for flattening two archetypes. That means understanding the
differential form of archetypes. That means understanding paths, node
ids and many other elements of archetypes. This is all primarily
described in the ADL2 spec
<http://www.openehr.org/releases/AM/latest/docs/ADL2/ADL2.html#_specialisation>
because that is the easiest way to comprehend it. Some elements are
described in the AOM2 spec, but it's harder to see, since now we are
talking in-memory object structures defined by class models.
So I remain convinced that languages have an important role to play in
our design, learning and understanding of things. Others may disagree ;)
- thomas
On 18/05/2017 16:22, Pablo Pazos wrote:
I really believe we should be teaching using tools not reading syntax,
specially for clinical modelers. If we are doing that right now is
because tools lack usability, features and maturity.
For techies, we like to look at the syntax because we need to parse
and process it.
I'm not against improving the syntax, but since we don't have much
resources as a community, shouldn't we focus were the real problem is
with tools instead of patching the specs?
Maybe clinical modelers can help software vendors on improving their
tools and to create new ones to help on the modeling process, and
there are some vendors creating such tools already but don't have
input from the community.
_______________________________________________
openEHR-technical mailing list
openEHR-technical@lists.openehr.org
http://lists.openehr.org/mailman/listinfo/openehr-technical_lists.openehr.org