diff options
author | nanbor <nanbor@ae88bc3d-4319-0410-8dbf-d08b4c9d3795> | 1998-10-27 09:28:34 +0000 |
---|---|---|
committer | nanbor <nanbor@ae88bc3d-4319-0410-8dbf-d08b4c9d3795> | 1998-10-27 09:28:34 +0000 |
commit | daa9c4033bdd041c50d24a52fee89135785ecbf1 (patch) | |
tree | 6e17e3f2d029b55359e1f56847270b862fbc8841 /ace/TP_Reactor.h | |
parent | d50ac6aaed74b028e0b12ed7014813874bb8ea75 (diff) | |
download | ATCD-daa9c4033bdd041c50d24a52fee89135785ecbf1.tar.gz |
Added a description section.
Diffstat (limited to 'ace/TP_Reactor.h')
-rw-r--r-- | ace/TP_Reactor.h | 16 |
1 files changed, 16 insertions, 0 deletions
diff --git a/ace/TP_Reactor.h b/ace/TP_Reactor.h index 12307f21d8e..3dab62c2607 100644 --- a/ace/TP_Reactor.h +++ b/ace/TP_Reactor.h @@ -9,8 +9,24 @@ // = FILENAME // TP_Reactor.h // +// = DESCRIPTION +// TP_Reactor (ala, Thread Pool Reactor) uses the leader-follower +// model to demultiplex requests among a bunch of threads. +// Basically, when using thread pool reactor, one will pre-spawn a +// _fixed_ number of threads. When you issue the run_event_loop () +// method, one of the thread will become the leader thread and wait +// for an event. The other threads (followers) will be lined up +// and waiting for their turns to become the leader. When an event +// occurs, the leader will pick a follower to become the leader and +// go on to handle the event. The consequence of using TP_Reactor +// is the amortization of the costs used to creating threads. The +// context switching cost will also reduce. More over, the total +// resources used by threads are bounded (because there are only so +// many threads.) +// // = AUTHOR // Irfan Pyarali +// Nanbor Wang // // ============================================================================ |