UUIDs
This page is outdated and no longer receives updates!
As of version OpenMRS v1.5 most objects/tables have a UUID property. This property is automatically filled in for every row and is 99.9% guaranteed to be unique across the universe. The primary reason for uuids is to support the Sync Module and allow for identifying objects that are identical across different OpenMRS installations.
In it's canonical form, a uuid value will be hexadecimal digits in the form of 8chars-4-4-4-12chars (e.g 0febc204-bca1-11de-913d-0010c6dffd0f). Depending on whether mysql or java generates these (see below), the uuid could be in version 1 or version 4 respectively.
The uuid can really be any unique string within the table, but to follow conventions, all OpenMRS uuids should be 36 characters in length and contain no whitespace. (Ideally, all OpenMRS uuids would be in the canonical form, but the uuids in the MVP dictionary, which were created manually, don't follow the 8chars-4-4-4-12chars convention, though they are all 36 characters in length.) Certain modules, such as the HTML Form Entry module, operate under the assumption that a uuid is 36 characters long and contains no whitespace, so if you are manually creating uuids for some reason, you should follow this convention.
All uuid columns can hold up to 38 character to accommodate MSSQL server's desire to put a curly brace ({}) at the beginning and end of the 36 chars.
You may hear the term GUID used as a synonym of UUID. GUID is Microsoft's implementation of UUID.
Primary Key Argument
The UUID is used as a normal property of every object instead of a primary key for several reasons:
Using long string uuids reduces performance. Its much faster to use smaller integers for PKs
The uuid is a long ugly 36 character string that would not look good in urls
Initial Column Values
When upgrading a current installation of openmrs, all rows will get a uuid generated for them. If you are using mysql, then openmrs uses the uuid() method during the update to do something like:
UPDATE obs SET uuid = uuid() WHERE uuid is null;
If you are using a DB other than mysql, most likely it also has a function to create a uuid. If not, then there is a java routine that runs and uses the java implementation of UUID generation. This is considerably slower.
There have been issues with mysql on windows and not getting unique uuids [1] . To find those duplicates, you can run this sql:
UUIDs and the OpenMRS API
Also in OpenMRS 1.5 there came a number of base abstract classes to define common metadata on objects. One of these objects was the OpenmrsObject. If an object extends this class and gets passed through an API method of the form of "save*(OpenmrsObject obj)", then the uuid property is automatically generated and saved along with the object. See API Save Handlers for more information about that magic.
The API uses the Java UUID class randomUUID() generation method (and so will be UUID v4 reference above).
OpenMRS UUIDs
There are some UUIDs shared across all OpenMRS implementations for metadata that are common across all implementations.
Concept Datatype UUIDs
Name | UUID |
---|---|
Numeric | |
Coded | |
Text | |
N/A | |
Document | |
Date | |
Time | |
Datetime | |
Boolean | |
Rule | |
Structured Numeric | |
Complex | |
Role UUIDs
Role | UUID |
---|---|
Anonyous | |
Authenticated | |
Provider | |
System Developer | |
HL7 Source UUIDs
Name | UUID |
---|---|
LOCAL | |
Location UUIDs
Name | UUID |
---|---|
Unknown Location | |
Patient Identifier UUIDs
Name | UUID |
---|---|
OpenMRS Identifier | |
Old Identification Number | |
Concept Class Identifier UUIDs
Name | UUID |
---|---|
Test | |
Procedure | |
Drug | |
Diagnosis | |
Finding | |
Anatomy | |
Question | |
LabSet | |
MedSet | |
ConvSet | |