Conversation

Notices

1. AndStatus (andstatus@loadaverage.org)'s status on Tuesday, 08-Nov-2016 20:00:22 UTC AndStatus

   • thefaico
   • Annah
   • Eugen 🎄
   @maiyannah @gargron On duplicated posts please read this discussion: https://github.com/andstatus/andstatus/issues/83
   • AndStatus (andstatus@loadaverage.org)'s status on Wednesday, 09-Nov-2016 05:35:56 UTC AndStatus

     • thefaico
     • Annah
     • Eugen 🎄
     @maiyannah As written in the above mentioned discussion, I think this behaviour of the server would be much better, even better than Twitter does ;-):
     ---
     Meanwhile I checked how Twitter responses to posting the message ("tweet") with duplicated text. It returns HTTP error "Forbidden", even if the duplicated message is not the last message sent by this user.
     It will be far more intelligent to respond not with an error but as a successful sending of that previous message. 
     Thus a client application (and a User of that application! ) will know that the message it's trying to resend was actually sent earlier. And stop resending!
     ---
     See https://github.com/andstatus/andstatus/issues/83
     What do you think?
     @gargron
   • AndStatus (andstatus@loadaverage.org)'s status on Wednesday, 09-Nov-2016 05:57:00 UTC AndStatus

     • thefaico
     • Annah
     • Eugen 🎄
     @maiyannah There could be many reasons, why a client application resends a message, not only its own error ;-)
     Another quote from the old discussion:
     ---
     In this case, I guess, we need to decide on the expected behavior of AndStatus first.
     My intention is to ensure that messages, posted via AndStatus, will be actually delivered to the server. Hence it is expected that in a case of an error while sending a message, AndStatus will retry sending the message later (up to 10 times for now).
     There may be different errors during sending, including errors sending delivery confirmation (actually, synchronous response with a posted message) back from the server to the client resulting in re-sending of the message that was already successfully delivered to the server (because AndStatus doesn't know that the message was successfully delivered). This is a normal situation and it should be handled by communicating systems...
     Do you agree so far?
     ---
     @gargron
   • AndStatus (andstatus@loadaverage.org)'s status on Wednesday, 09-Nov-2016 08:12:57 UTC AndStatus

     • thefaico
     • Annah
     @maiyannah@community.highlandarrow.com I agree, and as I wrote, AndStatus resends only those posts, which were not successfully delivered to the server from the application's point of view.
   • AndStatus (andstatus@loadaverage.org)'s status on Wednesday, 09-Nov-2016 10:27:00 UTC AndStatus

     • thefaico
     • Annah
     @maiyannah The point of view is incorrect, but it doesn't mean that this is a client's error or bug! And this why I proposed an intelligent fix at a server side...
   • AndStatus (andstatus@loadaverage.org)'s status on Wednesday, 09-Nov-2016 10:48:54 UTC AndStatus

     • thefaico
     • Annah
     • Eugen 🎄
     @maiyannah I think that relying on a client application to check for duplication is conceptually wrong, because this solution won't guarantee absence of duplicated posts at a server side. E.g. a client application reposts intentionally, or has a bug...
     @gargron