meta edit and rule based reasoning

Sure, quite a lot of customers have some kind of mapping DSL. We also have cases that generate Prolog facts, which are then processed by the customer's rules to identify the best solution and output it. In the cases I know, the output is some kind of code or data file, not something that is imported back into MetaEdit+.

Getting the output into MetaEdit+ is no problem - you can either output in the MXM Model XML format and import that, or then have Prolog or another program call the MetaEdit+ API to modify the existing models or create new ones.

Feeding the results back into MetaEdit+ to create models may however leave you open to the old problem of editing generated results, rather than the original source. Obviously there's no way to ensure the original model and the transformed results model remain in sync - unless the transformation is a simple bijection, which sounds unlikely if you're using Prolog.

You say you want the results fed back into MetaEdit+ "to produce reports". Maybe you can get the desired report straight out of Prolog? It's maybe no harder than having Prolog produce the results in MXM format. Alternatively, maybe the result from Prolog is small and simple, and the final report is something that needs to combine that with the original models. Depending on the balance between the two parts, you could either have Prolog do the combining (it presumably has most of the model info already, and any extra bits could be included in the original generation from MetaEdit+), or then you could have Prolog write its part of the results in a simple text file, and then read that with MERL and combine it with the model data:

1. generate Prolog data
2. run Prolog rules on data with external..executeBlocking,
   producing a text file, e.g. in .ini-style, key=value
3. read Prolog result into variable $PrologResult:
        variable 'PrologResult' write
           filename 'PrologResult.txt' read
        close
4. iterate over model and $PrologResult, parsing with regex translators:
      foreach .Input
      {   id '>'
          do $PrologResult
          {   if id%key = id;1 then id%value endif
          }
          newline
      }

The translators use regular expressions to identify and output the left and right sides of the .ini mapping:
   to '%key'   newline '/([^=]+)=(.*)/ $$1' endto
   to '%value' newline '/([^=]+)=(.*)/ $$2' endto

If there's a lot of data, you can improve the O(M*N) performance of step 4 to O(M*logN) by parsing the $PrologResult lines into variables first:

   do $PrologResult { variable id%key write id%value close }
   foreach .Input { id  '>'   variable id%key read   newline }