## Notice: This repository is no longer maintained as of 3/26/2024
Amazon S3 wheelhouse generator.
Wheels are the latest standard in distributing binary for Python. Wheels cut down scipy's installation time from 15 minutes to 15 seconds. Learn more about wheels.
Generate wheels for all listed PACKAGEs and their dependencies,
then upload them to Amazon S3 BUCKET:
$ mkwheelhouse [options] BUCKET [PACKAGE...] [pip-options]Then install with pip like usual, but preferring generated wheels:
$ pip install --find-links BUCKET/index.html PACKAGEYou can also build a wheelhouse in an S3 subdirectory by specifying the full S3 path:
$ mkwheelhouse s3://BUCKET/SUB/DIRECTORY PACKAGE-h,--helpPrint usage information and exit.
-e,--exclude WHEEL_FILENAMEDon't upload built wheel with filename WHEEL_FILENAME. Note this is the final wheel filename, like
argparse-1.3.0-py2.py3-none-any.whl, not the bare package name.Specifying an exclusion will not remove pre-existing built wheels from S3; you'll have to remove those wheels from the bucket manually.
-a,--acl POLICYApply canned ACL policy POLICY to the uploaded wheels and index. Specifying
public-read, for example, will make the uploaded wheels and index publicly accessible, avoiding the need for for a bucket policy to do the same. Valid values for POLICY can be found in the AWS documentation.
Any unrecognized options are passed directly to pip wheel. That
means mkwheelhouse supports requirements files, for example:
$ mkwheelhouse s3://BUCKET -r requirements.txtNeed an install log? Need to compile from source? You got it.
$ mkwheelhouse s3://BUCKET --log log.txt --no-binary :all: numpyConsult pip wheel for a complete list of pip-options.
- Python 2 and 3
- Set a bucket policy to make all objects publicly accessible or Pip won't be able to download wheels from your wheelhouse.