Thank you very much for the suggestion/idea

I've been thinking about this and how to best proceed with it. I do not have the requirement for this feature - but I would not be against adding it. However - I won't maintain this feature und thus would need to kindly ask if you can then be the person maintaining this feature. Additionally to that, it only makes sense for me to accept this feature, if we can achieve a high enough coverage that any changes to other code paths would at least show that this feature is broken - I do hope you understand this.

In regards to the design of the feature - my major concern would be how to keep the mapping safe between exporter restarts? Is there a good way to avoid the requirement of maintaining persistence?

I won't maintain this feature und thus would need to kindly ask if you can then be the person maintaining this feature.

Sure when there is a bug/feature request regarding pseudonymization just ping me.

Additionally to that, it only makes sense for me to accept this feature, if we can achieve a high enough coverage that any changes to other code paths would at least show that this feature is broken.

I never do PRs without providing tests. In my understanding you do something for me when accepting code and code without tests is always a burden for the project maintainer. So for me this is just common sense :-)

my major concern would be how to keep the mapping safe between exporter restarts?

As far as I am concerned we don't. By "pseudonymization" I meant "pseudonymized between restarts, anonymized across restarts". For me this is a feature and not a bug because by analyzing a pseudonym for a longer timespan might help identifying the real person/account. I only need to perform short-term analysis like finding out if one client uses a lot of bandwidth.

I hope this answers all your qustions :-)

@patrickjahns Any comment on this?

@phihos

Finally got around to thinking through it - let's move forward with it and document the limitation for that feature 👍