possible duplication of telemetry events

[edublancas]

we recently ran a workshop, and most attendees ran the code locally. then, we saw an uptick of events in posthog that didn't match the number of attendees: we had about 20 attendees but saw >100 new users on posthog. I suspect this has to do with #15, but I'm unsure.

the workshop covers a few of our projects, but I suspect the problem is with the Running experiments in parallel section, which runs this pipeline.

We need to investigate if this is still an issue and fix it.

FYI: @idomic

[yafimvo]

Yes, It seems like the problem happens when we use the parallel executor and we check if it's the 1st usage.

def check_first_time_usage():
    """
    The function checks for first time usage if the conf file exists and the
    uid file doesn't exist.
    """
    internal = Internal()
    first_time = internal.first_time
    internal.first_time = False
    return first_time

Since the default values of the Internal config file are these:

class Internal(Config):
    """
    Internal file to store settings (not intended to be modified by the
    user)
    """
    last_version_check: datetime.datetime = None
    uid: str
    first_time: bool = True

    def uid_default(self):
        return str(uuid4())

If we use parallel, and each process sees an empty config file, it gets the default values and creates a new user id. Since we're using internal = Internal() a lot of times this happens often.

I implemented a solution in this PR

I tried to reduce the number of times we write to this config file and changed the check_first_time_usage logic to read directly from the configuration file.

I ran the workshop locally and it seems to solve this duplications issue and didn't affect the tests.
This is my user: b76b83ed-8075-4e5b-ad8c-766aa4e773e9 (~186 events)