@pervognsen "Want" is a strong word here, though. Not being able to dequeue an IOCP because someone else did isn't an error, you just loop.
-
-
Replying to @cmuratori
@cmuratori So you'd loop an WFMO on an IOCP then try to do a zero timeout GetQueuedCompletionStatus?2 replies 0 retweets 0 likes -
Replying to @pervognsen
@pervognsen@cmuratori Honestly, I think the right approach is to decouple concurrency throttling from the IO completion port itself.1 reply 0 retweets 0 likes -
Replying to @pervognsen
@pervognsen@cmuratori "Uniform thread wakeup mechanism" is one of those things a new OS designed now could get right, I think.1 reply 1 retweet 1 like -
Replying to @rygorous
@pervognsen@cmuratori NT got a lot closer than what exists in Unix, but it's still full of annoying corner cases :/1 reply 1 retweet 0 likes -
Replying to @rygorous
@rygorous@cmuratori Yeah. Wonder what the right trade-off for NT is. IO completion ports with wait-and-forward threads for low freq events?2 replies 0 retweets 0 likes -
Replying to @pervognsen
@pervognsen@rygorous I don't know much about the NT scheduler internals, so it's hard to say one way or the other, but...1 reply 0 retweets 0 likes -
Replying to @cmuratori
@cmuratori@pervognsen@rygorous ... literally all you need is an GetQueuedCompletionStatus that is capable of returning one OOB signal.1 reply 0 retweets 0 likes -
Replying to @cmuratori
@cmuratori@pervognsen@rygorous That's all you would need, because 99% of the time you are just on the (one) IOCP.1 reply 0 retweets 0 likes -
Replying to @cmuratori
@cmuratori@pervognsen@rygorous And in the rare case that it returns the OOB signal, you can then go do whatever you want.1 reply 0 retweets 0 likes
@cmuratori @pervognsen @rygorous But there are probably better solutions if you were designing the whole thing again from scratch, etc.
Loading seems to be taking a while.
Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.