Lithops

Added
- [Core] New way to create RabbitMQ resources

Changed
- [Core] Default invoker background processes set to 2
- [Core] Code refactoring

Fixed
- [Core] Fixed issue when config in runtime is used multiple times
- [Core] Fixed invoker stop() method
- [Core] Some other Internal fixes

Added
- OpenWhisk Compute backend
- openwhisk_executor()
- Allowed multiple users in same CF namespace
- Added IBM COS request retrying when ReadTimeoutError

Changed
- COS token will expire 10 minutes before
- CF IAM token will expire 10 minutes before
- Improved remote invoker
- Reraise exception from functions
- Docs updated
- Default runtime timeout set to seconds
- default function timeout set to 595 secs

Fixed
- Fixed new invoker usage in notebooks
- fixes in knative backend
- Some other Internal fixes

Added
- New invoker mechanism
- New native remote invoker for ibm_cf
- pywren-runtime clean command to delete all tmp data
- capacity to limit the number of concurrent workers
- architecture documentation

Changed
- Changed Internal data cleaner logic to delete only desired job
- Updated ibm_cf Dockerfiles
- Moved chunk min size from 1MB to 0MB
- changed executor id format
- Timeout waiting for functions to complete set to None by default
- Updated ibm_cf base image requirements

Fixed
- Internal fixes
- Fixed tests
- Fixed pywren inside pywren function executions

Added
- New local_executor() to run pywren jobs in the local machine
- New localhost compute backend
- New localhost storage backend
- New docker_executor() to run pywren jobs in the local machine by using docker
- New docker compute backend

Changed
- Docs updated
- Code refactor

Fixed
- Internal fixes
- Bump pillow from 5.4.1 to 6.2.0

Added
- Allowed partitioner to split files by a number of chunks
- Missing logic in knative backend

Changed
- Docs updated

Fixed
- Internal fixes

Added
- Added knative-serving compute backend
- Added Dockerfile skeleton for slim Python3.6 runtime (only 307MB)
- Added CACHE_DIR in ~/.pywren/cache
- knative_executor() and function_executor()
- support to work on multiple regions at a time

Changed
- Docs updated
- Runtime Dockerfiles updated
- Runtime requirements updated
- Updated Cloudpickle lib to version 1.2.2
- Parameters introduced in the executer now overwrite the config
- updated tests

Fixed
- Internal logic to generate runtime_metadata
- invalid call to "is_remote_cluster" method
- Cloudpickle lib to accept any kind of import
- include_modules option in serializer