This isn't getting pushed until AmandaC's tested it.
I'm not sure this is such a good idea, anyway ; if it's for load/save
workflow improvements, wouldn't it be better to have a file access method
that allows for file re-opening?
That said, there's a limit to *that* before you just have to say,
"Just use /data/".
It's clinging to within an inch of its life, but the 192K save/load is working,
now I've cleaned up the critical path.
Any requests that'll bloat critical APIs are going to have to be terminated,
though, because this is getting out of hand.
The things I have to destroy, the joy in it!
For all I remove the more I save, the better it is!
Perfection is when there is nothing to take away, and I revel in this task.
Since this is after the technical "release", version numbers have been bumped to 1.
Changes before this commit for R1:
Kernel memory usage reduction schemes, with some security fixes.
Still need to deal w/ proxies (see later)
Changes in this commit:
Some various little things in apps
CLAW inet actually works now on 192K
sys-icecap no longer uses the event/neoux combination,
and now handles Everest disappearance as a mass-close,
but still handles Everest not being around on window create.
So it still handles every situation that matters.
neoux no longer handles everest crash protection.
Security policy and filedialog obviously don't use neoux anymore.
Kernel now only guarantees parsing, not event-loop, by executeAsync
This is safer and allows app-launcher to get rid of NeoUX by
any means necessary.
wrapMeta cache now exists, and proxies get wrapMeta'd to deal with
various low-priority security shenanigans.
This is a *stopgap* until I work out how to force OCEmu to give me
totally accurate boot-time memory figures, so I can create the
ultimate lowmem proxy. I'm calling it "puppet". FG knows why.
