Add telemetry

[valentinsulzer]

Description

Adds very basic telemetry - records when a simulation is solved.

The following section has been added to the user guide to explain the telemetry we are doing:

PyBaMM collects anonymous usage data to help improve the library. This telemetry is enabled by default but can be easily disabled. Here's what you need to know:

• What is collected: Basic usage information like PyBaMM version, Python version, and which functions are run.
• Why: To understand how PyBaMM is used and prioritize development efforts.
• Opt-out: To disable telemetry, set the environment variable PYBAMM_OPTOUT_TELEMETRY=true or use pybamm.telemetry.disable() in your code.
• Privacy: No personal information (name, email, etc) or sensitive information (parameter values, simulation results) is ever collected.

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 99.42%. Comparing base (444ecc1) to head (2e222c6).
Report is 1 commits behind head on develop.

Additional details and impacted files

@@           Coverage Diff            @@
##           develop    #4441   +/-   ##
========================================
  Coverage    99.41%   99.42%           
========================================
  Files          293      295    +2     
  Lines        22554    22599   +45     
========================================
+ Hits         22423    22468   +45     
  Misses         131      131           

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[agriyakhetarpal]

We could also make pybamm.telemetry.disable() a top-level pytest fixture as a cleaner way to disable telemetry when running tests

[valentinsulzer]

We could also make pybamm.telemetry.disable() a top-level pytest fixture as a cleaner way to disable telemetry when running tests

Yes, but we also want to disable it when our dependencies run their tests, so that wouldn't be enough right?

[agriyakhetarpal]

We could also make pybamm.telemetry.disable() a top-level pytest fixture as a cleaner way to disable telemetry when running tests

Yes, but we also want to disable it when our dependencies run their tests, so that wouldn't be enough right?

Oh, yes, forgot about that! In that case, we would need both the logic you added and the fixture to call it.

[kratman]

I understand why we want to know which parts of PyBaMM our users are using, but I am not sure this is the best approach.

My concerns:

• I feel like I would be upset to find out a free research tool was tracking my usage and sending it to a server. I think for a paid tool the expectation that they want to know your usage is a little more reasonable.
• Users installing through pip might not realize we are tracking their usage. It would be better to allow an opt-in instead so that users know we are collecting their data.
• Are we only going to be tracking simulation data so that we can just keep it in the simulation class or are we going to have to add the tracking to a lot of different places and make sure we add it to new functionality?

[martinjrobins]

Thanks @valentinsulzer. I suggest a timeout with default opt-out so we don't break ppl's CI, see below

[martinjrobins]

this should time-out after a short period with default no so we don't break everyone's automated stuff (say 10-20 sec?)

[kratman]

Yeah I agree, we should not have an infinite loop here

[agriyakhetarpal]

A better option could be to check for the ci_env_vars above (can be set as a constant at the top of the file) and assume non-interactive if set, and return early (like nox does it with session.interactive).

[valentinsulzer]

This function is only ever called by generate which already returns early if it detects an automated environment, but I'll also add the time check as a backup

[martinjrobins]

Suggested change

	if user_input in ["yes", "y"]:
	if user_input in ["yes", "y", ""]:

So if user presses enter it will be interpreted as a yes (as per the prompt)