Fetch variables by name when evaluating a decision

[saig0]

Description

Currently, the DMN engine doesn't provide a way to fetch the variables lazily. Instead, all variables must be provided when evaluating the decision.

When the DMN engine supports the lazy fetching of variables by their name then we should use it in the DMN module to avoid unnecessary database access and variable transformations. This should look similar to the expression language module.

blocked by #8082

requires camunda/dmn-scala#106

[falko]

Could such a lazy fetching of variable allow for an extension point to load data from outside the engine, e.g. through a job that a worker or connector could implement? I'm asking because this is in discussion for DMN 1.6

[saig0]

Could such a lazy fetching of variable allow for an extension point to load data from outside the engine, e.g. through a job that a worker or connector could implement?

@falko no. This is not the intention of the issue. It's just about the internal integration of the DMN engine inside Zeebe.

We have the same concept for the FEEL engine. It accesses the Zeebe data and loads the required data. But we don't use the same concept for the DMN engine yet.