On Thu, May 24, 2012 at 11:58 AM, Carsten Haitzler <[email protected]> wrote: > On Wed, 23 May 2012 20:13:24 +0200 Leif Middelschulte > <[email protected]> said: > > it seems cedric loves to starve the workers. never let cedric be in charge > man... just think. you'll be starving. :) seriously - the thread workers did > this: > > while (queue_a) handle(queue_a); > while (queue_b) handle(queue_b); > > normal ecore-thread workers are on queue a... ecore-thread FEEDBACK workers > are on queue b. this means as your workers keep re-scheduling themselves in a > tight loop queue a is ALWAYS full and busy and queue b never gets a turn. > fileselector uses eio. eio uses both thread AND thread feeback workers. thus > its tasks for queue b were being held "pending" forever due to your > pathological re-scheduling workers. > > i've fixed this in svn now and now everyone gets a turn. but remember this. if > you care WHEN your thread is run - then ONLY use ecore-thread FEEDBACK workers > and set the "create a thread" flag to true. this says "i must really have a > thread for this and not make use of the thread pool like everyone else". > otherwise your thread goes into the thread pool to get an available worker > WHEN > one becomes available. that may happen at any time in the future - depending > on > how many things are in the queue.
if not already done, shouldn't this go into some doc / examples ? Vincent ------------------------------------------------------------------------------ Live Security Virtual Conference Exclusive live event will cover all the ways today's security and threat landscape has changed and how IT managers can respond. Discussions will include endpoint security, mobile security and the latest in malware threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/ _______________________________________________ enlightenment-devel mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/enlightenment-devel
