Space Usage for botocore python library

[caiotoledo-lunasystems]

I'm using amazon-cloudwatch-publisher to publish journal logs and metrics to CloudWatch in one of our embedded projects.

But I've noticed that botocore python library takes a lot of space in our system:

# du -sh /usr/lib/python3.10/site-packages/boto*
1.4M    /usr/lib/python3.10/site-packages/boto3
44.0K   /usr/lib/python3.10/site-packages/boto3-1.26.20.dist-info
83.6M   /usr/lib/python3.10/site-packages/botocore
240.0K  /usr/lib/python3.10/site-packages/botocore-1.29.21.dist-info

Most part of these 83.6 MB is from its data folder.

Looking at amazon-cloudwatch-publisher code, it doesn't use all libraries from botocore.

Would it be possible to split botocore into multiple packages so the applications which depend on it could use a subset of botocore data files? Therefore using less disk space.

[thomas-roos]

Hi, thank you for this - will investigate and keep you updated.

[caiotoledo-lunasystems]

Seems that there are discussions about reducing botocore data folder size here: boto/botocore#2365
In case this feature is implemented this issue will be solved.

[thomas-roos]

I think this will be solved when that feature is implemented. Hope this is done soon.