Cool idea. Seems like it would require an entirely new philosophy vs our present one on security.

Yes, but it also removes a lot of footguns. Access to resources (ie. paths mostly) is controlled almost entirely by the parent process, which makes access controls highly pluggable and flexible.

The real problem is Plan9 never really had a lot of attention put on the things that make having a sane security policy good. Factotum seems, at best, to be bolted on after the fact.

> Factotum seems, at best, to be bolted on after the fact.

What gives you this impression?

It literally was, it didn't exist until the 4th edition of Plan9. That isn't to say it isn't a good idea (or implementation), but security is very clearly not a primary concern in Plan9.

> but security is very clearly not a primary concern in Plan9.

That is a myth that keeps getting propagated. https://plan9.io/sys/doc/auth.html

That paper is about factotum which was introduced in 4th edition, like I said. Regardless, I'm more talking about the fact that transport encryption still isn't used ubiquitously to my knowledge.

> That paper is about factotum which was introduced in 4th edition, like I said.

Which describes that yes, there was security in Plan 9 prior to Factotum, just that it wasn't good enough.

> Regardless, I'm more talking about the fact that transport encryption still isn't used ubiquitously to my knowledge.

It certainly is. You get SSL/TLS for free on Plan 9 as its a service. You dont mess with security code and instead use tlssrv(8). See https://man.9front.org/8/tlssrv

I didn't see there wasn't, I said it wasn't a priority.

I stand corrected on tlssrv

Yes, you would eventually be capable of sharing GPU power, devices, audio, anything. Imagine all your machine´s idle power available to others. Right now your GPU is barely being used.