[Top][All Lists]

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: About fathers and sons

From: Thomas Bushnell BSG
Subject: Re: About fathers and sons
Date: 12 Mar 2005 15:19:17 -0800
User-agent: Gnus/5.09 (Gnus v5.9.0) Emacs/21.3

Jose Luis Alarcon Sanchez <> writes:

>   When a program is designed for do some things, it's logic suppose that
> it gonna do this things and no others. If i give you the instructions
> for cook "paella valenciana" and you get a cake... some was wrong. :)

Yes, but the semantics of the functions do not guarantee that the
operation of "fork" is deterministic in this way.  

>   What you say can be true if the execution of the binary was made 3 or
> 4 times. But after more than 50 times with exactly the same output i am
> sure that the reason of "the lost" of this two processes are into the
> program and not out, in another place of the system.

So the Hurd creates two processes.  Why do you care?  :)  The extra
one is doing something, no doubt.  Or it's just giving you odd numbers
because it wants to.  You *can't assume anything* about the nature of
process ID assignment.  The system is entirely free to assign them
however it wants.  At random, if it wants.

>   My friend k0ro suggest that the explanation of this processes behavior
> in the GNU/Hurd system can be caused by the procceses server and not by
> the gnumach kernel. Opinions, please?.

Hurd process IDs are assigned by the process server, but that doesn't
affect the rule across all Unix that the system is entirely free to
assign process IDs in any way it sees fit, and you can conclude
nothing about the system from the IDs it happens to pick.


reply via email to

[Prev in Thread] Current Thread [Next in Thread]