3 ms·
I'm curious about the battery and reliability effects of push notifications without FCM. Using a single service for notifications allows a single wakeup across
by inlined 8y ago
I'm curious about the battery and reliability effects of push notifications without FCM. Using a single service for notifications allows a single wakeup across the device when fetching notifications. Last I checked, the minimum interval for coalesced timers in android is 15m and that's just in time for most telco NATs in the US to have silently killed your connection.
[Source: I used to run a 3P push network on android]
- chrismartin 8y agoIn my totally subjective experience, the reduced power consumption of doing away with Google Spy Services exceed whatever background drain I've suffered with 3+ apps doing their own messaging for notifications. My phone (Moto G5 Plus) has been off the charger for over a day now and is still at 65% battery. As for reliability, I have notifications for email (using K9 Mail and IMAP idle), XMPP (Conversations), and Signal. Emails (using IMAP idle) always reach me within a second or so. Occasionally, Signal messages will take hours to reach me, but I suspect that's not a problem with the websocket connection to my phone, as others running Signal (on a Google-laden phone) have had the same issue.
- lucb1e 8y agoWell I'm not sure how Telegram does it, but I get messages pushed instantly, it doesn't use Google spy services for that, and it's super low in battery usage rankings. I might be wrong but isn't it simply about building a tcp connection whenever Internet is available or changes, and the server will push whenever there's something new? You don't need a single thing polling (which would indeed save power) if the server is pushing. Recently I learned that sleeping (power-conserving) clients are even part of the WiFi spec, so that an access point will hold onto messages for some time until the client is listening again. I assume this is just some ms, but still.