Support Bulk Mode in PythonCaller

Liz, what is bulk mode?

It's also called a feature table, but I'm not sure if Safe still uses that definition. More info here:

https://knowledge.safe.com/questions/49482/what-transformers-have-feature-table-technology-fo.html

Thanks, it dawned on me that would be it. Everything should work that way by default.

I think there are quite a few people using Python to read custom data into FME. If Feature Tables could be created using either a PythonCaller or via the FME Plugin-SDK for custom Readers, than many of these processes could perform way faster. I guess otherwise only the C++ SDK would make sense for larger datasets.

With emphasis on Machine Learning and all of the tools available in python to interact with those models, It would be really cool to have Feature Table support both directions (as input into python and output from python back into FME). Could be really powerful to read some obscure format with FME, throw it into python to run a Linear Regression or some other model on then get the results back to process further in FME.

Any news on this one? It would be great to be able to use directly Pandas dataframe or other "tables" when there is a huge dataset instead of writing to csv first..

As bulk mode is being added to python based formats, do you have any idea if it will be soon available for everyone through the API?

In Beta!