My preference is for the logs to live alongside their output data files rather than in a specific fixed directory. (e.g. see #286 ) For example, when we build our networks, I like to have the output folder structure look like this:

• network_PBA50_v02
  • 2015 = folder with 2015 networks
  • 2019 = folder with 2019 networks
  • 2022 = folder with 2019 networks
  • 2025 = folder with 2019 networks
  • (etc through horizon year)
  • network_build_20220401_1015_INFO.log = this includes a list of each project applied before each write; this file can be consulted to answer "what projects were included in the year XXXX network", etc.
  • network_build_20220401_1015_DEBUG.log = this includes debug details from the network building process, including validation issues that don't cause a hard failure in the network building process, but modelers might want to be aware of as a possible issue

So the location and name of the log is at the discretion of the script or notebook that created the log, but I tend to think they're most useful when sitting with the output for future forensic purposes.

Addressed in merged PR