Burke --

It's a reality only because there is no system admin function separate from the 
developer/implementer function so there is no one to stop it.  In the big shops 
where I have worked, there are separate development, training, production and 
in some cases staging instances.  No developer would ever be given access to 
the production DB (except as a normal user, such as a time and attendance 
system).

We get frequent questions here on how to run multiple instances of OpenMRS 
under Tomcat, and during  demos I've seen some of the core developers running 
multiple instances of OpenMRS on their machines.  That is certainly one 
solution to the testing/training DB issue, and if there are issues of file 
placement and naming or environment variables that inhibit the ability to run 
multiple instances on the same machine, we should correct them.  In the DHIS2 
installation I worked on in Ghana, we set up separate training and production 
instances.

However, the new tools seem to make it so easy to do the right thing, we should 
stop doing the wrong thing.   Then it won't be reality.

From: [email protected] [mailto:[email protected]] On Behalf Of 
Burke Mamlin
Sent: Thursday, March 29, 2012 5:08 PM
To: [email protected]
Subject: Re: [OPENMRS-IMPLEMENTERS] Labeling patients as test in a production 
instance of OpenMRS

Roger,

Whether or not it's good practice, it's reality.  Test patients are common 
within production systems for … er … testing (surprise!), training, 
troubleshooting, demonstrating, etc.  It's generally only an issue when 
building cohorts for reporting/research.  We just need those tools have an easy 
way to recognize & filter out test patients (and anyone pulling data out 
directly via SQL would need the list of test patients too, but that's common 
practice).

Pragmatically speaking, best practice, is to moderate the creation of test 
patients in order to keep the list from continually growing and to make the 
patients easily recognized (e.g., with names like "TEST PATIENT").  At 
Regenstrief, we've used the identifiers 9-1, 99-2, 999-3, etc. for test 
patients because they're easy to remember (figuring out the check digit is 
easy) and easy to recognize.

Rather than having "test" status permeate the API like voided/retired, we could 
probably get what we need with two small changes that reporting & related tools 
could be refactored to use: (1) something like a 
CohortService.getTestPatients() method in the core API and (2)  a utility 
function to remove test patients from a Cohort .

Cheers,

-Burke
On Thu, Mar 29, 2012 at 4:22 PM, Friedman, Roger (CDC/CGH/DGHA) (CTR) 
<[email protected]<mailto:[email protected]>> wrote:
With the tools we now have (h2 DB, testing DB builder), does it really make 
sense to put test data in a production server?  It's certainly not best 
practice.

From: [email protected]<mailto:[email protected]> 
[mailto:[email protected]<mailto:[email protected]>] On Behalf Of 
Michael Seaton
Sent: Wednesday, March 28, 2012 10:06 PM
To: 
[email protected]<mailto:[email protected]>
Subject: Re: [OPENMRS-IMPLEMENTERS] Labeling patients as test in a production 
instance of OpenMRS

There is a pending ticket in the reporting module (REPORT-143 - Add option to 
exclude Test/Fake Patients from running 
queries/reports<https://tickets.openmrs.org/browse/REPORT-143>) for supporting 
this in the reporting module.

The planned design is to allow for a saved cohort definition to represent the 
cohort of test patients (this allows for test patients to be defined as a given 
implementation wants).  Then, the reporting module would exclude these patients 
from any query / report that it produces.

That being said, I wouldn't be at all opposed to some sort of "test" column on 
the person table...

Mike


On 03/28/2012 09:35 PM, Burke Mamlin wrote:
This has come up before.  While I can imagine a module trying to make test 
patients behave as if they're voided, we should probably make this a core 
feature (marking persons/patients as test).  As Ada points out, a person 
attribute can serve very well as a way of tagging test patients for exclusion 
from reports.

-Burke
On Wed, Mar 28, 2012 at 9:13 PM, Yeung, Ada K. 
<[email protected]<mailto:[email protected]>> wrote:
AMPATH creates a person_attribute_type of test patient.  Whenever we create new 
test patients, we tick the test patient person_attribute_type on the dashboard. 
 When it’s time to generate reports or prepare datasets for research studies, 
we can exclude those test patients easily.

-ada


From: [email protected]<mailto:[email protected]> 
[mailto:[email protected]<mailto:[email protected]>] On Behalf Of 
Joaquín Blaya
Sent: Wednesday, March 28, 2012 6:16 PM
To: 
[email protected]<mailto:[email protected]>
Subject: [OPENMRS-IMPLEMENTERS] Labeling patients as test in a production 
instance of OpenMRS

Hi,
Is there a way to have test patients in a production version of OpenMRS that 
allows them not to be counted in statistics e.g. some kind of label?

The idea is that in a production server you can have a handful of test patients 
and a test form so that when people start to learn it they can use those.

Thanks,

Joaquín
___________________________________________________________________
Gerente de Desarrollo, eHealth Systems<http://www.ehs.cl/>
Research Fellow, Escuela de Medicina de Harvard<http://hms.harvard.edu/>
Moderador, GHDOnline.org<http://www.ghdonline.org/>

________________________________
Click here to 
unsubscribe<mailto:[email protected]?body=SIGNOFF%20openmrs-implement-l>
 from OpenMRS Implementers' mailing list
________________________________
Click here to 
unsubscribe<mailto:[email protected]?body=SIGNOFF%20openmrs-implement-l>
 from OpenMRS Implementers' mailing list

________________________________
Click here to 
unsubscribe<mailto:[email protected]?body=SIGNOFF%20openmrs-implement-l>
 from OpenMRS Implementers' mailing list

Reply via email to