Github user stevedlawrence commented on the issue:
https://github.com/apache/nifi/pull/3130
Actually, I should probably add a little more clarification.
It is correct to say that Daffodil could be considered a transformation
capability, but I'm not sure it could really be seen as an alternative to Jolt
specifically.
One could definitely write a DFDL schema to describe and parse JSON data,
similar to what Jolt can do, but the resulting infoset would just be JSON that
is probably more generic and less useful that what Jolt would create. Daffodil
does not have the same transformational capabilities that Jolt has (e.g.
filtering, sorting, reordering). It really just takes a data format that is not
natively supported and transforms it to a logically similar representation in a
natively supported format (i.e. XML/JSON). If your data is already in something
like XML or JSON, it doesn't really make sense to use Daffodil.
So for example, you might take PCAP data, like in the above template, and
feed it through Daffodill to transform it to JSON. But there might be
extraneous data that Daffodil parsed that you don't need or care about. So you
might then feed that JSON through Jolt to to transform it to something that has
only the fields you care about, or is reordered/renamed to be something more
useful to ones usecase.
So this isn't something that could be viewed as an alternative to Jolt, but
more like something that could be used in conjunction with Jolt to allow one to
transform JSON data that wasn't originally JSON.
---
