Algo que me llamó la atención sobre el uso de select es lo siquiente:
Stevens(from memory) "when faced with the choice of using non-blocking i/o or threads/processes, the smarter decision is threads/procesesses." reasons:
*non blocking i/o is difficult to implement and error prone.
*the behavior of select with non blocking descriptors is platform specific. so you need added code to handle these differences
*threads/forks()'s are only a fraction of the time slower, and much easier to implement
Esto lo tomé de un foro (http://forums.devshed.com/t93193/s.html) en donde recomienda el siguiente libro:
Unix network programming volume 1 by w richard stevens
En fin, hasta el momento parece ser que he conseguido mi propósito, ahora faltará observar algunas cuestiones de rendimiento.
ILA
No comments:
Post a Comment