[Important] We started requiring concept descriptions, but this was a mistake

Continuing the conversation from

Background

Over the years it has been suggested several times that we should require that every concept have at least one description in some locale. Theoretically this would be the right way to go, but practically this would be a huge pain for implementations (especially existing ones), and among other things it makes many concepts in the CIEL dictionary invalid. Based on implementer feedback, we decided not to make this change.

Detailed history

Here are two tickets about this that we closed as Won’t Fix: TRUNK-3616, and TRUNK-3478. And here is the discussion where this decision was made.

However it looks like there was another ticket that we did not close, created in 2011 and finally implemented as part of platform 2.0: TRUNK-2809.

Situation Today

As a result starting with Platform 2.0 we require at least one description for every concept, and because of this “plenty” of CIEL concepts break. I.e. Platform 2.0 does not support the CIEL dictionary. (Or alternately, the CIEL dictionary does not support Platform 2.0.)

My proposal

I propose we revert this change in Platform 2.0.1, and go back to our prior behavior, where it’s allowed to have concepts with no descriptions.

The alternative to this is that the CIEL dictionary team has to add a description for every concept. (@akanter can say whether this is even an option.)

Regardless, I consider this a blocker for Reference Application 2.5 – we cannot release a refapp version that’s incompatible with the full CIEL dictionary.

1 Like

Agree. It is not appropriate to require a “description” or definition with each concept. Reference mappings provide definitions in most cases.

Agree with the proposal.

I have been blocked before by this feature. Surely many concepts have no descriptions and making it a requirement, is a show stopper for many existing implementations :+1:

+1000 this is a blocker and needs to be reverted/relaxed…

I agree this should be reverted. I still think we should be promoting best practices – i.e., striving to have descriptions for anything we are promoting as a dictionary concept in OpenMRS.

FYI – I reopened TRUNK-2809 and made it ready for work to be reverted and closed as Won’t Fix. I also tagged it as community-priority.

TRUNK-2809 has been closed as a won’t fix with the changes, of requiring a concept description, reverted.

In regards to this, should the reference application 2.5 depend on platform 2.0.1?

I think so, yes.

That does introduce a new dependency though…

-Darius (by phone)