Hello. You are right. If it works and you are satisfied, you should keep your solution. By the wya JoltTransformation may be difficult at the very beginning. But it is very powerful and with some pratice, it begins to be easy.
For study, you may give it a try. Regards. Etienne Jouvin Le jeu. 5 déc. 2019 à 14:40, James McMahon <[email protected]> a écrit : > Hello Etienne. Yes, Matt may have mentioned that approach and I started to > look into it. > > My initial thought was this: is it much of a savings? My rudimentary > process works in three process steps - each simple in configuration. The > JoltTransformationJSON would eliminate only one processor, and it looks > fairly complex to configure. It appears to require a Custom Transformation > Class Name, a Custom Module Directory, and a Jolt Specification. For folks > who have done it before those may be an afterthought. But as is often the > case with NiFi, if you've never used a processor sometimes it is hard to > find concrete examples to configure NiFi processors, services, schemas, etc > etc. I opted to take the more familiar path, not being familiar with the > Jolt transformation processor. > > Am happy to learn and will see if there's much out there in way of > examples to configure JoltTransformationJSON. For now I'll use my less > elegant solution that works gets me where i need to be: pumping data > through my production system. > > Good suggestion. Thanks again. > > On Thu, Dec 5, 2019 at 8:20 AM Etienne Jouvin <[email protected]> > wrote: > >> Hello. >> >> Why don't you use a JoltTransformation process first to produce multiple >> element in JSON according value in the array, and duplicate common >> attributes for all. >> And then, you do the split. >> >> Etienne >> >> >> >>
