On 24 May 2017, at 20:34, Daniel Gultsch <dan...@gultsch.de> wrote: > > https://mail.jabber.org/pipermail/standards/2016-February/030925.html > > The follow up message from Holger (Timeout only starts after first > push-worthy stanza) is also important. > > We also came to the conclusion that it is desirable to actively close > the TCP connection if an <r/> from the server remains unanswered after > ~30 seconds or so. This is due to some platforms (read Android) not > allowing us to properly close the connection ourself. (But this is > probably a business rule for SM and not for Push). It also handled in > that module. > > Both the prosody module as well as the unpublished ejabberd follow > those business rules.
Thanks Daniel. I don’t (only) want to be contrary, but I would like these to be at most suggestions in the XEP (I’m not opposed to including them for the sake of server devs who want guidance, but I don’t think they should be normative). When I was thinking about this the other day I came up with a somewhat different scheme, which I’d like to also be allowable (I see no reason it shouldn’t be, given that it doesn’t cause interop problems). In related news, I think there’s value in extending this (optionally seems fine) to support retractions too. If a server knows that all the unread messages that it’s sent notifications for have now been read, it seems reasonable for it to be able to tell the XMPP Push Service to tell the device that. /K _______________________________________________ Standards mailing list Info: https://mail.jabber.org/mailman/listinfo/standards Unsubscribe: standards-unsubscr...@xmpp.org _______________________________________________
