oship-dev team mailing list archive
-
oship-dev team
-
Mailing list archive
-
Message #01318
Re: openEHR Refactoring
2010/8/6 Tim Cook <timothywayne.cook@xxxxxxxxx>
> On Fri, 2010-08-06 at 13:42 -0300, Diego Manhães Pinheiro wrote:
> >
> > These modules was generated by the adl2py code generator, correct ?
>
> Yes.
>
> > So, why they are necessary to the OSHIP, if they can be generated any
> > time you want ?
>
> Right. I deleted them for now in my code.
>
> > They aren't used by any other module and they don't have tests.
>
> I see some OSHIP concepts are missing here. See below.
>
> > I didn't catch how important is these files to the oship system. Could
> > you explain me ? I didn't catch their purpose .
>
> I'll attempt an overall explanation below.
>
> >
> > This approach changes how I understand what is the OSHIP project.
>
> That is obviously the case.
>
> > At the moment, OSHIP project is only and just only the implementation
> > of the openEHR model.
>
> This is incorrect. You are *only* working on that portion right now.
>
Ok.
>
> > Until that I know, the purpose of all openEHR components is mix all
> > concepts from different models (hl7, ISO 18308, CEN) into a model that
> > communicate with all definitons described by each different
> > specification. My point is if separate them means that OSHIP approach
> > is leaved aside.
> > Could you explain me this too ?
>
> Okay. OSHIP == Open Source Healthcare Information PLATFORM
>
> In this context a PLATFORM == a solid base upon which interoperable
> applications can be constructed.
>
> Also see: http://dictionary.reference.com/browse/platform
>
> Starting from the assumption that OSHIP will help create applications more
flexible and more interoperable, what difference from the framework meaning
? So, is it correct to say that we are creating a framework ? I'd like know
your opinion about it.
> So OSHIP is and always has been defined as an application development
> platform. The ultimate goal is to have an infrastructure where
> information from multiple places based on multiple models can be brought
> together or disseminated from.
>
> Even if it were only going to be openEHR though, you will note in the
> openEHR documentation that the top-level namespace is 'openehr'. This
> is easily seen in the ADL documentation where the naming of archetypes
> is discussed.
>
> We cannot know exactly what applications will be built on OSHIP. This
> is a key concept to the overall project.
>
> Now to revisit those other modules.
>
> The km (knowledge models) branch is used to contain the python
> representations of the constraint models. The reason we want to do this
> is for application speed at runtime. If the ADL (openEHR) or XSD
> (MLHIM) models have to be parsed each time at runtime this is very
> inefficient. If they are converted one time to python the each time an
> instance needs to be created it is much faster. ADL and XSD files are
> used to transport the same knowledge models to different applications on
> different platforms in different programming languages.
>
I personally don't agree with this approach. I think If generate code, means
that
anyway you have to maintain it. If you have to maintain it, the goal of
reduce the
complexity is lost.
A insight : A constraint engine with a persistent configuration could also
solve
without lose performance.
>
> Think about having a data entry screen with maybe 50 - 100 constraint
> definitions. Then they may (likely will) have dependencies on each
> other at runtime as well.
>
>
> > I take your point, but there are some points that need be clear before
> > this merge be done.
>
> I hope I have answered your questions. But eitherway; the openEHR
> top-level namespace is 'openehr' not 'rm'. You have openehr.rm and
> openehr.am
>
> Thanks for your explanations, Tim. :)
Coming back to actual problem, now I got it why the openehr namespace must
be kept.
Do you mind change they using a new revision? I finished all conversions
applied to the openehr modules, but using the wrong way.
Just moving directories and using a simple regular expression I can change
it so fast.
> Cheers,
> Tim
>
>
>
>
--
http://dmpinheiro.net
-----------------------------------
Diego Manhães Pinheiro
Follow ups
References