[ On Sunday, December 5, 1999 at 07:10:29 (GMT), Christos Zoulas wrote: ]
> Subject: Re: A report on implementing runlevels in NetBSD
>
> Oh, yeah? Every new AIX version needs tweaks to get the job control
> code working in tcsh. I have no idea why. Seem to be changing the
> process group stuff. Or to build you need to #define _BSD_INCLUDES
> or -lbsd or, or, or...

I don't recall there being any explicit separate BSD API on AIX, at
least not in 3.x.x (I've not really done much programming on 4.x, and
what I did was actually running on an Apple Server and using GCC :-).

I think I did port what there was of zsh and pdksh to AIX-3 back then,
but I don't recall any problems with job control.

> And the beauty of using an *environment* variable to choose between
> overcommiting and not overcommitting malloc.

Vmalloc uses environment variables to control its behaviour too.

Or are you talking about the virtual memory algorithm and SIGDANGER?

As I recall there was no way originally to control which processes got
SIGDANGER and I do remember my X session being killed because the X
server was the largest single process running when things got tight....
Having some way to say that the X server is the last process that should
get SIGDANGER was definitely necessary.

-- 
							Greg A. Woods

+1 416 218-0098      VE3TCP      <gwoods@acm.org>      <robohack!woods>
Planix, Inc. <woods@planix.com>; Secrets of the Weird <woods@weird.com>