How well devs in haiku knows Plan 9 From Bell-Labs? Have you been imitating it? ( please answer yes. :] )
I liked Plan9 implementation, to what I learned about it. It all looked very simple in principles. Inceribly simple! I didn’t know one would be able to leave THAT much useles things away from operating system. It’s closeness to perfection blinds me.
I’ve not yet been able to learn haiku better, partly because of it’s beta state. It’ll be intresting to compare plan9 and haiku.
Plan9 is more like ‘for developers’ -OS, thought, in it’s brilliancy I’d be suprised if it didn’t go well for an usual user as well.
from what I understand about Plan9, the goals of the project were quite different from what BeOS, and by extension, Haiku were/are out to achieve. Plan9 seeks to bring distributed computing to the table, and has strong roots (for various reasons) in UNIX. BeOS/Haiku were/are out to make the user the central point around which the OS serves. On that very superficial level, the two projects appear to be working towards seperate goals.
However, Plan9 is a lot of different things. What did you have in mind that Haiku devs should consider, and why?
Well, they do have different goal, but both ends up being more positive than usually. And Plan9 ends up being something very alike what haiku has a goal. The distributed computing is not only thing what plan9 makes easier. It fixes "everything is a file" -philosophy and provides private namespaces for processes. I think it is much more, power, control, pathway to real easy, than just putting user into the central point in which OS serves. I mean, it is obvious computers serve users. The best part is how does they serve, and Plan9 gives some sweet dreams about it.
Plan9 looks so simple comparex to unix, still implementing incredibly many things. To simplicity: Counting the user interface, source code and way how it is reached.
I don’t really understand what you are trying to get at, but this could be due to the language.
Plan9 is an OS designed for distributed networks. While there are users at the end of this grid somewhere, that’s not quite the same as the user in the Haiku sense of the word. Haiku is, at least as far as I understand the rhetoric so far, an OS designed for a user, on a machine, being presented with a UI that enables them to excel in lots of media based operations and tasks. Not run a distributed grid powering lots of thin clients.
Plan 9 is not just for distributed systems. I’ve run it on a non-networked machine before. It has a lot of good idea; it manages to save a lot of Unix’s better features, but improving on it’s weak points. BeOS/Haiku are similar in this respect.
the only plan 9 I know of … is the one from outer space
"Plan 9 is an operating system kernel but also a collection of accompanying software. The bulk of the software is predominantly new, written for Plan 9 rather than ported from Unix or other systems."
According to their site the 4th edition (latest) was released in 2002 and there seems to be something called Inferno, which you can buy … looks like it was built on Plan 9 ???
It has some interesting concepts and is light weight, but that’s where the comparison to BeOS/Haiku proly stops.
It has some interesting concepts and is light weight, but that's where the comparison to BeOS/Haiku proly stops.
You did read that somewhere, didn’t you? Because I’ve noticed they had some kind of similarity. I otherwise wouldn’t started this thread.
I like both things in Plan9 and haiku, it seems they have taken opposite paths to reach more perfect solutions, but both of them ended up with something I happen to like.
expensivelesbian wrote:
I don't really understand what you are trying to get at, but this could be due to the language.
I’m trying to reach some developer in haiku team to understand and accept an idea about imitating Plan9 to make haiku R2 or R3 take benefits from Bell-Lab’s important work which provided us with Plan9. It’d be hard, but I believe it’s worth of it to try. Just do not fall into trap just imitating features from Plan9’s surface, it’d be failure (and exactly what people in MS seems to do).
to make a statement like “we should explore the design concepts of <OS_here>”, one must at least provide some good reasons why. So far, I’ve seen, Plan9-is-good-let’s-do-it-for-Haiku! That’s great, but please, which design concepts would be good for Haiku, and why? The “why” bit is perhaps the most crucial, as it’s all to easy to copy and paste the features list of a wikipedia article and leave it at that.
A bit more flesh to the bare bones of your topic, would be awfully peachy.
Nah, I don’t think this needs any more flesh. I could mention program namespaces, rio, that nice manner to get rid from sockets and handle totally everything as a file. Thought I’d keep it as a grand failure to just implement them into haiku. I believe it’s the perfection what counts, and I just propose ppl going learning how they could possibly do some things better. I’m not proposing turning haiku to plan9, I’m proposing learning from good work and research, to do even better. I think your last post was ridiculous, I thought I was already made it clear.
Things like the union directories and remote devices were discussed a long while back on the glass elevator list (Though as I recall there was no real conclusion other than ‘Hey wouldn’t this be cool if…’).
The way Plan9 goes about doing these things would be quite interesting to add to Haiku. Having, for example, my wife’s printer or her PDA appear as connected directly to my computer over a LAN would be awesome; It could potentially simplify the way in which we interract with such devices.
As for union directories, they are yet another awesome filesystem trick which I’d support (Along with the rest of the FS bag of tricks discussed in GE at some point in time).
I’d be wary of putting too much into the filesystem though, the idea of literally everything being a file appears a little over the top to me though I could still be convinced.
Yes, the mention of devices and Plan9’s approach to how they are employed on a network was something I always appreciated, especially because it seemed so…obvious that that’s how things should be. In that respect, yeah, it was very “Be.”
I’m trying to think how distributed computing could be the next step after SMP. Imagine games that can use the cpu power of thousands of idle computers across the net.
I'm trying to think how distributed computing could be the next step after SMP. Imagine games that can use the cpu power of thousands of idle computers across the net.
only possible if we all have very very high speed reliable net connection. I think in the future, GPU’s do all visual processing on a PC (much like high end workstations).
It’s closeness to perfection blinds me.