Release 0.56.3

Fixed license fields.

Release 0.56.2

Fixed CI/CD to fix PyPI release process.

Release 0.56.1

Bug / developer experience fix: We now raise a RuntimeError if the user attempts to call Plugin.publish outside of a with block.

Release 0.56.0

Improve accuracy of Camera image timestamps by time stamping between frame grab and decode steps.

Contributed by @gemblerz.

Release 0.55.1

This release drops support for publishing bytes values via Plugin.publish.

Though this was technically allowed by pywaggle before, the node data pipeline would silently drop messages with bytes data leading to confusion on for the end user.

In the future, we may change our position on this but our current recommendation is to either:

1. Unpack binary data into individual int, float or string fields.
2. Manage your own serialization and deserialization scheme.

Release 0.55.0

This release updates the run log data.ndjson format to be compatible with the data API and sage-data-client's load function.

This should reduce tooling friction when transitioning from development to deployment of apps.

Release 0.54.0

Graduated run log directory out of alpha and added usage documentation to docs.

Release 0.53.5

Bug fix: Actually fixed subscribe. (Fixed internal queue passed to consumer.)

Release 0.53.4

Bug fix: Fixed plugin subscribe. (Consumer was not providing routing key correctly.)

Release 0.54.0 Alpha 1

Initial alpha test of pywaggle log directory support!

If the environment variable PYWAGGLE_LOG_DIR=path/to/dir is defined, then pywaggle will publish measurements and uploads to path/to/dir as plain files in the following format:

path/to/dir/
    data.ndjson
    uploads/
        nstimestamp1-filename1
        nstimestamp2-filename2
        ...

Messages are currently logged in data.ndjson in the following "raw", line separated JSON format:

{"name":"env.temperature","ts":1656449073603869000,"meta":{"sensor":"bme280"},"val":23.0}
{"name":"upload","ts":1656449073603976000,"meta":{"camera":"top","filename":"test.png.webp"},"val":"/Users/sean/git/pywaggle-log-dir-example/testrun/uploads/1656449073603976000-test.png.webp"}
{"name":"image.cats","ts":1656449074613277000,"meta":{},"val":0}
{"name":"image.birds","ts":1656449074613507000,"meta":{"camera":"left"},"val":8}
{"name":"timeit.inference","ts":1656449074613580000,"meta":{},"val":1005408000}

The log is flushed for each measurement, so piping the log and commands like tail -f path/to/dir/data.ndjson work as expected.