News for hackers. Stuff that matters.
Re:Rob Pike on Systems Software Research (2000) by jacquesm on Saturday January 30, 2010 @08:35PM (#1089754)
You probably should add a (2000) to the title.

I would really like to see Rob Pike's updated version of this.

He's right about the 'meat' of it though.

I thought that by now we'd be using some kind of ultra stable micro kernel based operating system, using message passing as the primary interface between components.

Instead, all we got was endless API calls.

Re:Rob Pike on Systems Software Research (2000) by glhaynes on Sunday January 31, 2010 @02:18AM (#1090200)
I assume the complaint is that there's not sufficient processor-assisted memory protection? In your opinion, which modules tend to interfere with other modules in today's OSs in a way that would be improved by this?
Re:Rob Pike on Systems Software Research (2000) by jacquesm on Sunday January 31, 2010 @05:40AM (#1090365)
Which ones do you'll never know, but potentially all of them.

Having a driver segfault is very convenient, it allows you to partition 'blame' in a way that you actually get somewhere.