Subject: Re: siginfo, signal queues, memory issues 
In-reply-to: Your message of "Mon, 02 Dec 2002 22:24:39 CST."
             <20021203042439.GB39614@NewGold.NET> 
--------


>So essentially you don't want reliable signals, right?

Sometimes, that's appropriate. When I was driving a device from
userspace code, and mapping device interrupts onto signals, and I'm
getting circa 30,000 interrupts a second (this was on a 120MHz pre-MMC
Pentium), then .... no, I didnt want reliable signals.  I takes 'em as
I gets 'em, cause I'm never going to keep up.

This isn't facetious; I did just that, for a very early implementation
of what's now called DDP or RDMA. ( See ``A caching model of operating
system and functionality, D.R. Cheriton and K. Duda, for one
exposition of the ideas.)