4 ms·

All modern OSs were designed in the 80s, with limitations and mind set shaped by the hardware limitations. Since then there was no major rethink of the OS arch

All modern OSs were designed in the 80s, with limitations and mind set shaped by the hardware limitations. Since then there was no major rethink of the OS architecture, I think.

hailwren 7y ago
This is just wrong. Operating Systems are an area of extensive ongoing research. New and wonky OS's are coming out all the time (Hurd, nix, Qubes, etc). Even macOS itself is being overhauled pretty frequently (i.e. APFS, whole disk encryption, etc etc)
- felipelemos 7y ago
  Hurd is still "under development" since 1990, so it came out 30 years ago. Not something that I would call new.
- java-man 7y ago
  Sorry, I should have said "mainstream OSs", meaning Linux, Mac OS X, Windows. Case in point - though not a topic of this particular discussion, neither of the three were designed with security in mind. Most of the security features are, in my opinion, bolt-on afterthoughts. This is just my opinion though.
- zadkey 7y ago
  Yes operating systems are in active development and research. For major companies that are profit-driven such as Apple or Microsoft, operating system development takes a major backseat as these companies have already carved out a sizable market share and don't need to invest much to maintain their dominance as they have self-sustaining app ecosystems for their OS which continue to draw users reliably. Sure there are people innovating everyday in operating system development like with Haiku and Redox. Many Linux and UNIX distros are very active as well. But for the major players, it's low priority, as it gives low return to shareholders. Which means the overwhelming majority of users won't benefit from most of these new innovations until they become sufficiently innovative that they become incredibly enticing (likely to expand the market share of a major player) or become either a market threat to a major player by being enticing to app developers leading to a them developing their own self-sustaining ecosystem of apps which draws users away from major players thus creating meaningful competition.
  - hailwren 7y ago
    This is still false. As a concrete example, see the timeline for MacOS ASLR rollout. The fact that you think things like Haiku or Redox are revolutionary and you aren't talking about kernel locking schemes, file systems, protection against control flow redirection or really any subject area that is considered part of active OS research makes me think that the reason you don't think big companies are working on OS's is because you aren't actually interested in OS research. Which is fine, there's no reason you should be. But it means you're barking up the wrong tree.
    - zadkey 7y ago
      Again, I wasn't saying there aren't innovations. But I would argue that for a company like Apple, their next iPhone release takes priority and will get a significantly larger allocation of their resources than the next OS level feature.
      - hailwren 7y ago
        Yes, I agree Apple spends much more on iPhone development than OS research. However, I'd be willing to bet that they pay their OS devs more. Further, the reason Apple gets to spend less on OS research is that everyone is working on OS research. OS research is at every institution and everyone benefits from OS advances (like ASLR from earlier or NX bits). Further, academic institutions are actively involved in OS research in a way that they aren't involved in helping Apple produce the next iPhone. So, sure -- Apple isn't investing as much, but they are keeping their OS modern and there isn't some corporate conspiracy (or even competitive opportunity) to keep users away from having Rust in the kernel. Overall, I think OS research is probably at a more interesting point in history than mobile hardware. There really is amazing competition happening between Linux/macOS/Windows and the users are getting those benefits. Research into secure computation and encryption is being integrated at the hardware level, security research is delving into lots of interesting problems w.r.t. badly behaving programs and also programs that behave correctly but attempt to perform malicious actions (i.e. buffer overflows vs keyloggers), and as lame as it sounds Linux's massive kernel improvements (I'm thinking specifically about their work on mutexes) have allowed them to become one of the quickest OS's in history. It also brushes up against the question of what we're even calling OS research. The secure enclave was an iPhone first piece of hardware, but its concepts, development, and execution were certainly done by people who I would call OS researchers.
saagarjha 7y ago
I can pretty much "update" my Linux computer without turning it off. Android can also do this I think.
- iudqnolq 7y ago
  Android can't

All modern OSs were designed in the 80s, with limitations and mind set shaped by the hardware limitations. Since then there was no major rethink of the OS architecture, I think.

hailwren 7y ago

This is just wrong. Operating Systems are an area of extensive ongoing research. New and wonky OS's are coming out all the time (Hurd, nix, Qubes, etc). Even macOS itself is being overhauled pretty frequently (i.e. APFS, whole disk encryption, etc etc)

felipelemos 7y ago
Hurd is still "under development" since 1990, so it came out 30 years ago. Not something that I would call new.
java-man 7y ago
Sorry, I should have said "mainstream OSs", meaning Linux, Mac OS X, Windows. Case in point - though not a topic of this particular discussion, neither of the three were designed with security in mind. Most of the security features are, in my opinion, bolt-on afterthoughts. This is just my opinion though.
zadkey 7y ago
Yes operating systems are in active development and research. For major companies that are profit-driven such as Apple or Microsoft, operating system development takes a major backseat as these companies have already carved out a sizable market share and don't need to invest much to maintain their dominance as they have self-sustaining app ecosystems for their OS which continue to draw users reliably. Sure there are people innovating everyday in operating system development like with Haiku and Redox. Many Linux and UNIX distros are very active as well. But for the major players, it's low priority, as it gives low return to shareholders. Which means the overwhelming majority of users won't benefit from most of these new innovations until they become sufficiently innovative that they become incredibly enticing (likely to expand the market share of a major player) or become either a market threat to a major player by being enticing to app developers leading to a them developing their own self-sustaining ecosystem of apps which draws users away from major players thus creating meaningful competition.
- hailwren 7y ago
  This is still false. As a concrete example, see the timeline for MacOS ASLR rollout. The fact that you think things like Haiku or Redox are revolutionary and you aren't talking about kernel locking schemes, file systems, protection against control flow redirection or really any subject area that is considered part of active OS research makes me think that the reason you don't think big companies are working on OS's is because you aren't actually interested in OS research. Which is fine, there's no reason you should be. But it means you're barking up the wrong tree.
  - zadkey 7y ago
    Again, I wasn't saying there aren't innovations. But I would argue that for a company like Apple, their next iPhone release takes priority and will get a significantly larger allocation of their resources than the next OS level feature.
    - hailwren 7y ago
      Yes, I agree Apple spends much more on iPhone development than OS research. However, I'd be willing to bet that they pay their OS devs more. Further, the reason Apple gets to spend less on OS research is that everyone is working on OS research. OS research is at every institution and everyone benefits from OS advances (like ASLR from earlier or NX bits). Further, academic institutions are actively involved in OS research in a way that they aren't involved in helping Apple produce the next iPhone. So, sure -- Apple isn't investing as much, but they are keeping their OS modern and there isn't some corporate conspiracy (or even competitive opportunity) to keep users away from having Rust in the kernel. Overall, I think OS research is probably at a more interesting point in history than mobile hardware. There really is amazing competition happening between Linux/macOS/Windows and the users are getting those benefits. Research into secure computation and encryption is being integrated at the hardware level, security research is delving into lots of interesting problems w.r.t. badly behaving programs and also programs that behave correctly but attempt to perform malicious actions (i.e. buffer overflows vs keyloggers), and as lame as it sounds Linux's massive kernel improvements (I'm thinking specifically about their work on mutexes) have allowed them to become one of the quickest OS's in history. It also brushes up against the question of what we're even calling OS research. The secure enclave was an iPhone first piece of hardware, but its concepts, development, and execution were certainly done by people who I would call OS researchers.

saagarjha 7y ago

I can pretty much "update" my Linux computer without turning it off. Android can also do this I think.

iudqnolq 7y ago
Android can't