my company is interested in using a customized Cammunda DMN modeler (from bpmn.io) with the Cammunda BPM service backend. We succeeded to realize deployment with subsequent evaluation but would also like to provide functionality for testing during table design without the need to deploy first - similar to the testing capabilities in https://dmn.camunda.cloud/.
We couldn’t figure out if this is possible and which operations would be required, looking at the service documentation. Using the cloud service operation POST /evaluate is not an option because our customers usually work in environments where access to the Internet is restricted. Using the Java library to implement an evaluation service is also not attractive since Java skills are not sufficiently present.
Research in this forum and also the old forum didn’t lead to an answer. In a similar topic (Redirecting to Google Groups) the solution was based the Java API. So we decided to post that question.
Is there a service in Cammunda BPM, hosted on premises, for testing a DMN table without the need for a previous deployment, similar to POST /evaluate of the cloud solution?
thanks for the quick response. We already have the Camunda platform available and we are using the evaluate operation. This operation however requires you to deploy prior to evaluation - all paths either require a key or id - whereas the cloud version of evaluation allows you to pass the XML along with the parameters.
We would like to avoid the need of deploying a table only for testing purpose since we fear this has impliciations and undesireable sideeffects (cleaning up after evaluation, cleaning up after an unexpected crash/node failure, visibility of temporary tables not intended to be used in production, performance).
and then return the result of the calls back to the user.
From this perspective users would only need to make a call with the deployment and variables.
Sure, that’s also something we are aware of. Yet we percieve this approach as not failsave and eventually with sideeffects (see sideeffects we expect in my previous comment). We need to expect failure of hardware or infrastructure all the time.
Basically I like to find out if there is a single, side-effect free service operation available in the REST API overlooked by us. We had this situation before so it helps to ask sometimes. If not we would consider the alternatives.
Also if somebody has arguments against the sideeffects mentioned above, this would of course also be valuable.