jdev - 2023-10-18

a note about that should be added to the XEP

Hi XMPP devs. I would like to show you my analysis of Android messaging apps: power draw, bandwidth and RAM use. I intend to publish this more widely over time. But for now I feel it's a great moment to seek comments from developers. https://decentim.grafana.net/public-dashboards/92602d3a4aa842ce97812d310077691d

wow, nice

yaxim is so low, it doesn't even make the stats 🤭😄

andrey.utkin: great overview!

Ge0rG, indeed, non-installed apps need the lowest amount of energy!

andrey.utkin: very interesting, thanks for sharing (and for concluding that XMPP _the_ permacomputing-friendly choice ^^)

Can you add "time spent on the app" graphs too? I may be wrong, but I think this would be a fairer comparison to divide everything by the time the app was "active" on the devices? If you send 100 whatsapp messages per day but only 1 XMPP message every other day, then it's quite important for those stats. The "Narration" part gives some nice insight already, and it's pretty nice to test on idle devices. Maybe the "daily driver" stats should be separated from the "idle" ones? And about the "daily driver" test devices, this is where it would be super nice to have an indication of time spent/messages exchanges/number of contacts/…? Maybe you don't want to do that for obvious privacy reasons though :)

The "Median measurements per app", no matter how much I love seeing Conversations at the first place of the podium in every category is not fair. Example: WhatsApp is only used on a "daily driver" device, but Conversations is used in 3 "idle phones" and 2 "daily driver" ones.

It's all OK since you also provide "Median measurements per device and app" though. Anyway, thanks!

nicoco, thanks for your appreciation! You are right on all accounts. But there's not enough stuff yet to separate the categories more. More devices are needed, stricter, fair synthetic conditions, and that will be a lot more of work and time. Measurements of human-used devices would become palatable if a lot of people volunteer their data, then some convergence of measurements may emerge. I will look for "app's screen time", maybe there's such thing in batterystats report, that would be great, but I have doubts.

Whatif aXeth'MPeePee is bridging to one of them bestest apps like Facebook (Meta) Messenger? How would the graph look‽

andrey.utkin: if you explain how to collect and send data, I’m happy to contribute. I only have one phone, with Cheogram (Conversations fork). I use my phone very little (I’m on my human-sized computer all day, why would I use my lilliputian-sized computer?), but I receive a fair amount of messages via XMPP. Let me know if you’re interested 😊

nicoco_, for how to send data, start with reading https://github.com/decent-im/messengers-perf

nicoco_, of course your data contribution is very welcome