Pywaggle

Fixed data shim timestamp to use time_ns to match plugin.

Added cv2 warning instead of crashing when importing data shim.

Image data publish assumes data will be uint8 [0,255] RGB pixel data. Any rescaling will be left up to user.

Internal plugin update to support better testing and publishing binary and image data publishing.

Release Notes

This is a _major_ update. In this release, plugins have been completely redesigned with the following changes:

1. Plugins are intended to safely instrument a whole program. The old interface has been simplified and can now safely instrument existing code. For example:


import waggle.plugin as plugin

init once for entire program. all publishing can now be called safely from anywhere
plugin.init()

...

while True:
plugin.publish('env.humidity.htu21d', 80.1)


2. Plugins can subscribe to data streams.
3. Plugins use ontology based names instead of explicit internal IDs.

The implementation of the Plugin class now runs in a more decoupled way from RabbitMQ. This will allow user code to run more predictably and to more robustly manage its connection to RabbitMQ in the background. In the future, this will allow cleaner management of things like RabbitMQ heartbeats, receiving messages, etc.

1. To address running the plugin with less required parameters, the waggle.plugin.Plugin() class uses the following default values:


WAGGLE_PLUGIN_INSTANCE = 0
WAGGLE_NODE_ID = 0000000000000000
WAGGLE_SUB_ID = 0000000000000000
WAGGLE_PLUGIN_USERNAME = worker
WAGGLE_PLUGIN_PASSWORD = worker
WAGGLE_PLUGIN_HOST = rabbitmq
WAGGLE_PLUGIN_PORT = 5672


2. Encoding and decoding signed integers and floats have been implemented in the protocol.