< Previous by Date | Date Index | Next by Date > |
Thread Index |
Guys,A previous commit of mine on trunk resulted a block condition in DialogUsageManager::process().
The cause was a Fifo::getNext(0), which would be blocked until there is a message. I've re-commited (after the revert from scott) my previous patch but used 1 instead of 0 as the timeout.
But I am not sure if this is the right semantics. The question is, is Fifo::getNext(0) supposed to be blocking?
On the branch I am working on mostly, Fifo::getNext(0) is a nonblocking call if it's empty. I failed to find useful log (due to re-org) to find out when and why the semantics diverted.
Nash,The revert you did was unnecessary, since Fifo::getNext(0) wouldn't block on our branch.
Ken