Community
Participate
Working Groups
Expected behaviour: Use the ETL trace as a model for further processing (i.e. store, print, etc.) via modelling scripts. Seen behaviour: Currently, the ETL trace model is a POJO with a trace cache and a list of generated artefacts per rule. Proposal: Do something similar to EPL in which the Match model implements IModel and provides a set of script based types: one type per match rule (rule XXX -> PatternXXX) and one type per rule role (role YYY -> PatternXXXYYY). For ETL I would suggest only one type per rule (rule XXX -> TxXXX), and then have the "transform" and "to" elements as properties. Perhaps a type for "transform" roles would be interesting too (TxXXXYYY), so we could provide an equivalent/equivalents operation (via the model).