Set local_cache_dir from environment variable?

[indigoviolet]

If I want a persistent cache, I can set CLOUPATHLIB_FILE_CACHE_MODE. But I then have to pass a local_cache_dir in code - why not allow this to also be set from the environment?

[pjbull]

Thanks @indigoviolet, good point. We should allow specifying a CLOUDPATHLIB_DEFAULT_CACHE_DIR.

We should check for this in in Client instantiation and use it if present. I think precedence should be:

• local_cache_dir if passed
• fallback to CLOUDPATHLIB_DEFAULT_CACHE_DIR if available
• fallback to TemporaryDirectory like we do now

Would be happy to accept a PR if you're so inclined. PR should include tests and docs.

[indigoviolet]

Thanks @pjbull, I will try to do that.

Another point I noticed was that there isn't a clean way to set up persistence/local_cache_dir while being agnostic to the type of path via AnyPath -> CloudPath. Do you have a suggestion for how to do that?

[pjbull]

Another point I noticed was that there isn't a clean way to set up persistence/local_cache_dir while being agnostic to the type of path via AnyPath -> CloudPath. Do you have a suggestion for how to do that?

Yep, that should be doable with set_as_default_client.

For example:

from cloudpathlib import AnyPath, S3Client

s3 = S3Client(local_cache_dir="./data")
s3.set_as_default_client()

cp = AnyPath("s3://test-bucket/file.txt")
cp.client._local_cache_dir
#> PosixPath('data')

Docs here:
https://cloudpathlib.drivendata.org/v0.15/api-reference/client/#cloudpathlib.client.Client.set_as_default_client

[pjbull]

This has beed added and shipped @indigoviolet. Try install 0.16.0 with pip if you want to experiment with it and confirm