Refactor Logging Hierarchy in neo4j-python-driver

[OliverFarren]

This PR includes a refactor of the logger names in the neo4j-python-driver to adhere to a hierarchical structure.

Moving to hierarchical loggers is a non-breaking changed designed to offer more control when managing logs and provides a mechanism for better organisation and filtering of logs.

import logging

logger = logging.getLogger("neo4j.async.io")
logger.setLevel(logging.CRITICAL)

[robsdedude]

Thank you for taking the time to craft a PR. Much appreciated.

However, before I can review the code itself, please sign the CLA (Contributor License Agreement).

General feedback from the PR description: I like the idea. However, I'd not want to leak the internal module structure. Instead, I'd prefer a logical hierarchy.

Something along the lines of

• neo4j
  • .network (socket related logging)
  • .bolt (protocol messages)
  • .pool (connection pooling, and probably also routing as those two concepts are inherently coupled)
  • possibly more?

[OliverFarren]

Thank you for taking the time to craft a PR. Much appreciated.

However, before I can review the code itself, please sign the CLA (Contributor License Agreement).

General feedback from the PR description: I like the idea. However, I'd not want to leak the internal module structure. Instead, I'd prefer a logical hierarchy.

Something along the lines of

• neo4j

  • .network (socket related logging)
  • .bolt (protocol messages)
  • .pool (connection pooling, and probably also routing as those two concepts are inherently coupled)
  • possibly more?

Hi @robsdedude, i've signed the CLA

That sounds good to me, so to clarify, you'd avoid the sync/async split?

[OliverFarren]

@robsdedude, i've consolidated the logger hierarchy, let me know your thoughts. database drivers are a bit outside my technical area so I don't have any input into the logical splitting.

[robsdedude]

Thanks again for the update. I went through the PR and adjusted it a little:

• add API docs

• merged network and bolt loggers back into one io logger (sorry for the extra work :grimacing:). Going over the code, I found it's really hard to draw a line between raw IO (there's not much being logged there) and the bolt protocol.

  Please have a look at the PR and let me know if it still looks good to you.

[OliverFarren]

@robsdedude could we get it merged in?