🚀 go-pugleaf

In article <1992Apr7.201330.10720@leland.Stanford.EDU>  
magnus@Kimura.Stanford.EDU (Magnus Nordborg) writes:
|> I fail to see why this is so awful. If you specify a "hiwat" mark so that
|> the swap file cannot fill the whole partition, you have basically turned
|> the swap file into a swap partition. We use a lot of swap on our system,
|> and on the SPARCs we need to set aside 300MB as a separate partition,
|> whereas on the NeXT, we can at least sometimes use that space...
Because the NeXT doesn't work the swapfile like a "normal" swap partition
when hiwat is set. On my slab programs will start dying when hiwat is hit,
as they should. BUT they start dying long before the swapfile is really
exhausted, ie swapfile hits hiwat but there is STILL ROOM IN THE SWAPFILE!
Worse, if I hit it the wrong way, my slab turns into a boat anchor and must be
recovered via pulling the power cord.

And the box is a real slug after it hits hiwat, if it survives to keep running
that is. You can kill everything, log out, log back in, and the slab is still
in slug mode. It doesn't seem to be very efficent at processing the swapfile
for free pages. Which isn't a surprise since it starts killing processes before
the swapfile is truly exhausted.

The swapfile is the single largest fault I've found with my slab. And as
luck would have it, it is a MAJOR fault in what would otherwise be a really
nice diamond. It is also *THE* reason that I haven't upgraded from my mono
slab. The more I try to live with this fault on my mono machine the more I
can't picture living with it on a ND or a color slab. Hey NeXT, do me,
and my local NeXT dealer, a favor and FIX THIS!!!

I mean I'm using a DOS box to review images processed on my slab because
viewing them on my slab is too damaging to the horsepower of my slab. And if
I had a color slab I would STILL be using the DOS box because of this swapfile
mess. So why buy a color NeXT?

Scotty