The pingback data has been incredibly informative, and now that it's usage has been established, I think we should add some more datapoints to it (aka pingback v2). This might warrant an RfC, I'm not sure. It also might be easier to break this down into smaller tasks. This came out of discussions with @Sj, @Addshore, and @CCicalese_WMF at Wikimania 2018.
- Whether specific extensions are installed.
- I don't think we really want a full extension list (especially since that could be used for identification), but whether specific extensions are installed. For example when gauging if a bundled extension has low usage and should be unbundled.
- Distribution mechanism (git, tarball, ...)
- This should be extendable so that distribution packagers should be able to modify the value (e.g. "fedora", "docker", "debian").
- Consider having two levels of opt-in, to report basic and advanced stats.
And relatedly:
- Make sure people can opt-in via the CLI installer, and then ask other redistributors to enable this (e.g. Meza)
Related: