Dear Hoppers,

The other day I had this idea which is just too cool not to act upon. The
idea is to provide methods of reflection for Hop.  In particular,
reflection on what's going on during the execution of pipelines and
workflows.  So to act upon these urges of mine I started working on a new
"Reflection" plugin for Hop.

The idea is to have a new type of metadata "Pipeline Log" which allows you
to specify to which pipeline you want pipeline logging information to be
sent at the start of any pipeline, at the end and periodically.  In the
specified pipeline you'll have a transform which will provide the actual
logging information in the form of fields.
We can then do the same for workflows of-course but also for data probes:
capture the data output of a specific transform, type of transform, all
transforms, ... and deliver it to another pipeline in a normalized
fashion.  You can do all sorts of cool things with it this way, I believe
in the realms of continuous data profiling, data quality, machine learning
and what not.

I guess it's one of these ideas that have always been around and sound so
incredibly simple but for some reason elude us.

Cheers,
Matt

Reply via email to