kuralabs / flowbber Goto Github PK

In some situations it is not relevant to store all the data from the bundle in the database.

Add a filter option to the databases sinks (currently MongoDB and InfluxDB) to allow to filter data to store.

Join all dictionaries keys (using the keysjoiner character) until a leaf value that has a datatype supported by InfluxDB (string, float, integer, or boolean) is found.

Add None as supported.

#16 wasn't passing in the CI because handler are now being installed in the root logger BEFORE configuring Flowbber's multiprocess logging.

Possible candidates of the issue are:

pytest-dev/pytest@3862b0b

Instead of flowbber[lcov_html]

One single collection is in some situations impractical.

Allow to set the collection to None and use the key of each entry in the data bundle to set the collection to use.

This will allow to pass configure parameters to the template engine.

This line is wrong https://github.com/kuralabs/flowbber/blob/master/lib/flowbber/plugins/sources/lcov.py#L263

As there are cases where lcov fails, yet the code is 0

$ lcov --rc lcov_branch_coverage=1 --directory /foo/ -capture --output-file /tmp/tmp8hlgxlg7
Capturing coverage data from /foo/
Found gcov version: 5.4.0
Scanning /foo/ for .gcda files ...
geninfo: WARNING: no .gcda files found in /foo/ - skipping!
Finished .info-file creation
$ echo $?
0

In order to allow setting NULL to a option in TOML, the user should be able to just remove the option.

For this, the default value of the option must be None.

This implies changes in default key for databases sinks (currently MongoDB and InfluxDB), which is better because it avoids having a dependency over the timestamp source. If they want to change it later to have a single timestamp id, they can add it later.

There should be an explicit warning that keys should be strings?