Bug 9243 - Subscribers should not be notified of requests that occur before they subscribe
: Subscribers should not be notified of requests that occur before they subscribe
Status: NEW
Product: Logitech Media Server
Classification: Unclassified
Component: CLI
: 7.2
: All All
: -- enhancement (vote)
: Future
Assigned To: Adrian Smith
:
Depends on:
Blocks: 9364
  Show dependency treegraph
 
Reported: 2008-08-21 08:31 UTC by Max Spicer
Modified: 2009-03-26 12:30 UTC (History)
0 users

See Also:
Category: ---


Attachments

Note You need to log in before you can comment on or make changes to this bug.
Description Max Spicer 2008-08-21 08:31:46 UTC
If you subscribe to a request using S::C::R::subscribe, you currently get notified of requests that executed _before_ you placed the subscription.  This is due to the asynchronous nature of notifications (and is probably also quantum).  It would be great if the notification mechanism could store a timestamp for when a subscriber subscribed and then only send notifications if the requests occurred after that.  If backwards compatibility is needed, the subscribe function could take an optional 'onlyFuture' parameter.
Comment 1 Adrian Smith 2008-09-23 14:23:29 UTC
I'll look at this - probably keep a queue of pending subscriptions and add when there's nothing in the notification queue.
Comment 2 Adrian Smith 2008-11-11 13:10:28 UTC
Max - afraid I won't get to this in time for 7.3
Comment 3 Adrian Smith 2008-12-31 05:19:02 UTC
Is this still needed Max?
Comment 4 James Richardson 2009-01-08 10:04:39 UTC
Moving to future till more feedback is received.  Please re target when feedback is received.
Comment 5 Simon Turner 2009-03-26 12:30:05 UTC
Not too sure if this is the sort of feedabck you meant James. Here's a personal scenario that will continue to happen until this bug is dealt with. I have anm alrm set up so that at 7.00pm every day my kitchen SB3 fires up into BBC Radio 4 so I don't miss any episodes from my favourite radio drama. It's 15 minutes long and when it's finished we may turn off the player, but most times we choose some music to lilsten to when it's finshed. At some later point this music then unexpectedly stops dead as the alarm has timed out.

If this bug is addressed then replacing the alarm playlist with a new one (as my family does) will cease the alarm function so the new playlist will continue to play (assuming it is implemented).

It is quite likely that this problem affects most people who set alarms for any other reason than to get up in the morning to go to work.