Supplying Naming Convention as Module Breaks Parallelization

[cwe5590]

dlt version

0.5.1

Describe the problem

On another note, re: custom naming conventions. I just wanted to flag that normalization parallelization fails when supplying a custom naming convention module directly to a destination. This is because modules can't be pickled for multiprocessing. Using the fully qualified path to the naming convention's module fixes this. I'm not sure if y'all consider this a bug or not, but could be helpful to document as it can be a bit tricky debug

Expected behavior

No response

Steps to reproduce

Supply a naming convention module to a destination and run enl steps using multiprocess

Operating system

Linux

Runtime environment

Local

Python version

3.11

dlt data source

No response

dlt destination

No response

Other deployment details

No response

Additional information

No response

[rudolfix]

@cwe5590 thanks for raising this! We'll mention that in our docs. Any workaround is a lot of effort and (IMO) not so much gain.