hi,
Ah, what a shame! I'll raise a feature request.
To call it a "debugger" is a bit of a stretch, it's effectively a replay of the evaluated model but it's good enough for now. To make it a "proper" debugger wouldn't be too hard I suppose...<thinking/>
Essentially what I've done is implement a DMNRuntimeEventListener which captures the before/after events and enhances the event information where possible. For example in afterInvokeBKM I capture the parameters & the values that were passed to the bkm.
I create a JsonObject representation of each event and associated data and add it to a simple List<JsonObject>. I post process this list to make it easier to work with (for example I collapse consecutive, related events such as before & after contextEntry into a single JsonObject). Then I dump the whole list to disk as json.
Finally, I have a simple (horrible!) little bootstrap app that lets me step through the serialised json list an object at a time and tries to represent the callstack, dependencies etc.
I was hoping to integrate this into the existing DMN editor so I could show the evaluated result next to each context entry.
If you'd still like to see it I'll happily share what I have once it's a bit more joined up.
Cheers,
Paul.