* De: Noriyuki Soda <soda@sra.co.jp> [ Data: 2002-12-01 ]
	[ Subjecte: Re: siginfo, signal queues, memory issues ]
> >>>>> On Sun, 1 Dec 2002 14:27:29 -0500,
> 	christos@zoulas.com (Christos Zoulas) said:
> 
> > Because it is a pain in the ass to code properly. If you make
> > two signal delivery paths, then how to you keep track of the
> > delivery order?
> 
> Well, the answer was already written in this thread. :-)
> 
> Please think about the following way:
> 
> - take only one signal delivery path. (i.e. queue)
> 
> - maintain number of queued signals for each signal
>   to limit the number.
> 
> - for non-realtime signals, you can limit the number to 1.
> 
> In this way, it's not pain to keep the order proper, because there is
> only one signal delivery path. Perhaps you may use a bitmap to count
> the number of signals for non-realtime signals, because the maximum
> number is 1.
> 
> I think it isn't better to make non-realtime signals queued
> even for SIGINFO case, because that should cut down the number of
> available global resource for realtime queued signal. And realtime
> signals are what we really need to make queued.

So essentially you don't want reliable signals, right?
-- 
Juli Mallett <jmallett@NewGold.NET>