Developing for openMRS 1.11.7

[johanneshk (Johannes Kitschke)]

Hello,

I’m working on a project where we want to deliver a module that should run on openMRS 1.11.7. Since I’m only getting to know openMRS, I have some questions and hope you can help a total noob get started with openMRS…

1. is the OpenMRS SDK 1.x the right place to start: https://wiki.openmrs.org/display/docs/OpenMRS+SDK+1.x ? Things I should know when developing for this version?
2. we want to integrate machine learning into our module, are there limitations (e.g. runtime, CPU, RAM) we should be aware of? The idea is to run the machine learning on a schedule (every week) and produce appointment-specific risk scores. What if this takes a lot of time and resources, can it somehow lock up openMRS?
3. is it possible to package the module and then distribute it via e.g. email so users can upload / install the module directly into openMRS?

Thank you!

Johannes

[ibacher (Ian Bacher)]

@johanneshk

is the OpenMRS SDK 1.x the right place to start: OpenMRS SDK 1.x - Documentation - OpenMRS Wiki ? Things I should know when developing for this version?

The latest version of the SDK should still work for developing even against old versions of the platform. Their versioning isn’t in any way connected to one another.

@johanneshk

are there limitations (e.g. runtime, CPU, RAM) we should be aware of

Resource-wise, it mostly depends on your hardware and JVM configuration. It is worth noting, though, that OpenMRS modules can really only execute Java code directly. And, yes, if you run machine learning models on the same server and take up all the CPUs, you could very easily bring OpenMRS to a halt. Are you planning on updating your models based on live data or simply applying them to live data? (In the later case, it’s very unlikely to eat up all the resources unless you have a very inefficient model… most of the time in machine learning, it’s the training and validation of the model that are computationally expensive).

@johanneshk

is it possible to package the module and then distribute it via e.g. email so users can upload / install the module directly into openMRS?

Yes… OpenMRS modules compile to OMOD files, which is basically a JAR containing the module code and its dependencies. You can easily email the OMOD around, assuming that it’s small enough that mail servers will allow it. Mostly whether or not you can mail it will depend on the resulting size of the OMOD which can vary greatly. For example, the FHIR2 module OMOD, which has numerous dependencies, currently is about 57mb, which is larger than most mail servers will allow. On the other hand, the SPA module, which is more or less self-contained, is only 13kb and could easily be emailed.

[herbert24 (Herbert Yiga)]

@ibacher

Yes… OpenMRS modules compile to OMOD files, which is basically a JAR containing the module code and its dependencies. You can easily email the OMOD around, assuming that it’s small enough that mail servers will allow it. Mostly whether or not you can mail it will depend on the resulting size of the OMOD which can vary greatly. For example, the FHIR2 module OMOD, which has numerous dependencies, currently is about 57mb, which is larger than most mail servers will allow. On the other hand, the SPA module, which is more or less self-contained, is only 13kb and could easily be emailed.

Thanks @ibacher ,@johanneshk just an add on,you can also upload the compiled omod on google drive and share it with any one