Build awsme for the Lambda environment #3
Description
- Amazon Web Services Cloud Watch Metrics library version: 0.2.5
- Python version: 3.6, 3.7
- Operating System: AWS Lambda
Description
The awsme library has as an install_includes boto3. This is perfect and appropriate for most applications of the library, but it does not work well for AWS Lambda applications of the library.
The reason is as follows: boto3 and the AWS API changes pretty rapidly, and AWS updates the Lambda runtime in sync with those changes. Best practices for Lambda functions are to NOT include boto3 in the lambda function package to prevent weird incompatibilities between the AWS API and older versions of boto3. Also, it prevents MAJOR package bloat.
What I've done in the past is fork project that have boto3, remove boto3 as a dependency, and upload the new build to pypi as lambda-. However, this is sub-optimal, obviously, as it only loosely tracks the primary project.
Could we alter the build to both produce awsme for standard applications and in the same run produce lambda-awsme for Lambda applications?