Scheme.org URL shortener Lassi Kortela (27 Jul 2022 14:17 UTC)
Re: Scheme.org URL shortener Lassi Kortela (05 Aug 2022 11:58 UTC)
Re: Scheme.org URL shortener Jakub T. Jankiewicz (05 Aug 2022 12:36 UTC)
Re: Scheme.org URL shortener Lassi Kortela (05 Aug 2022 14:08 UTC)
Re: Scheme.org URL shortener John Cowan (05 Aug 2022 18:07 UTC)
Re: Scheme.org URL shortener Lassi Kortela (05 Aug 2022 20:09 UTC)
Re: Scheme.org URL shortener John Cowan (05 Aug 2022 22:37 UTC)
Re: Scheme.org URL shortener Lassi Kortela (05 Aug 2022 22:59 UTC)
Re: Scheme.org URL shortener Jakub T. Jankiewicz (06 Aug 2022 11:09 UTC)
Log analysis for statistics Lassi Kortela (06 Aug 2022 14:29 UTC)
Re: Log analysis for statistics Jakub T. Jankiewicz (06 Aug 2022 19:13 UTC)
Re: Log analysis for statistics Lassi Kortela (07 Aug 2022 06:40 UTC)
Re: Log analysis for statistics Magnus Ahltorp (06 Aug 2022 21:37 UTC)
Re: Log analysis for statistics Lassi Kortela (07 Aug 2022 06:55 UTC)

Re: Log analysis for statistics Magnus Ahltorp 06 Aug 2022 21:37 UTC

> If I've understood correctly, Magnus allows us to crunch the server-side nginx logs to generate anonymous statistics. I just tried out the old Unix program `analog` to do that.

The important thing here is not the method of collection: reading through logs, javascript-based collection, or updating a database directly in the server-side code. The important thing is type of data and the dissemination. What is collected and who is it sent to.

On one end of the spectrum we have Google Analytics that collects as much as it can, and sends it into Google, and on the other end there is totally stateless servers (as Scheme users we should know a thing or two about side effects). In the EU, it is debatable if the Google Analytics end of the spectrum is even legal, but it is definitely a no-no for scheme.org. I would like us to be as close to the stateless end as possible, while still allowing things like a usage counter or other types of hopefully benign statistics.

(I don't think it's a good idea that I meddle too deeply in this, but a more privacy-preserving way to present the counters is to only update them infrequently, like once a day or once a week. This also scales when we have 1 million hits per second in the future.)

So, even logs is something that can be used for tracking, but if we rotate the logs frequently enough and only use it for debugging and very coarse statistics, that's fine.

We should really try to have proper written rules for this, and consensus is better than me occasionally answering emails. But the whole point of the current organisation of scheme.org is that Lassi gets things done, and I don't, so I wouldn't hold my breath waiting for me to come up with something.

/Magnus