20 Mar
2003
20 Mar
'03
7:50 a.m.
That sounds like a better approach. The only difference I can see is that the window for something else (outside the control of the pike I/O system) to read the data becomes greater, but that's a broken situation anyway.
/ Martin Stjernholm, Roxen IS
Previous text:
2003-03-20 08:39: Subject: I/O callbacks in blocking mode
I see another old suggestion is to use flags not to call read_callback if read was previously called on that socket but after the last poll.
Would that solve the thread problem? It seems simpler then using buffers.
/ Mirar