Cleaning up a few things I found when trying to run the evaluation demo on a fresh repo copy.
The only actual bugs were:
We needed to be creating the cache directory structure earlier; with ..., save_path.open("wb") as out_file otherwise errors on a fresh repository.
Use of the cache was being set to different default values in different places. I changed things to make the default to use it.
The other changes are a bit more stylistic and/or pedantic:
I replaced the contents of data/README.md with "see the help function" because we don't want to have multiple sources of truth (one is likely to end up out of sync otherwise).
I made the path to included-divisions.txt an argument in load_metadata with a default, rather than a strictly hard-coded value. In so doing, I decided the default place to call this from was the top level of the repo, rather than /data.
Cleaning up a few things I found when trying to run the evaluation demo on a fresh repo copy.
The only actual bugs were:
with ..., save_path.open("wb") as out_file
otherwise errors on a fresh repository.The other changes are a bit more stylistic and/or pedantic:
data/README.md
with "see the help function" because we don't want to have multiple sources of truth (one is likely to end up out of sync otherwise).included-divisions.txt
an argument inload_metadata
with a default, rather than a strictly hard-coded value. In so doing, I decided the default place to call this from was the top level of the repo, rather than/data
.