Plan 9: The way the future was

http://catb.org/~esr/writings/taoup/html/plan9.html

by niyazpk • 14 years ago

197 82 14 years ago

6ren• 14 years ago

Ugly now beats beautiful later.

An elegant solution requires a deep understanding of a problem, and that understanding takes a lot of data (awareness of cases), and then a lot of time and effort to create a simple theory of it and to create a simple solution embodying that theory.

bct• 14 years ago

The trick is to build ugly systems that are able to grow into beautiful ones.

jff5• 14 years ago

That is quite a trick. The problem is that pretty soon, everyone depends on the behavior of the ugly warts in your original system, so now you're stuck with them lest you break the holy Compatibility.

wladimir• 14 years ago

It seems that (almost?) everything that was planned out for plan9 is finding its way into operating systems.

The ideas were good, but somehow they had to take the long way around. It was way ahead of its time.

_delirium• 14 years ago

I remember an interview with one of the ex-Plan9 people now at Google (Rob Pike maybe?) talking about how he's switched to Linux, since that's what Google runs on, and finds it a strange experience, as if a bunch of bugs you thought you'd fixed 20 years ago have resurfaced in the main branch. But Linux does seem to be picking up a steady number of the features.

shimon_e2• 14 years ago

It was Rob Pike from the slashdot interview. http://interviews.slashdot.org/story/04/10/18/1153211/rob-pi...

pavpanchekha• 14 years ago

First, one should note that this is a chapter (or section) of "The Art of Unix Programming", by esr; it is a fantastic book (I'm going through it now, on chapter 9 of 20, and the wisdom has been near-palpable).

As for the specifics of the article. Plan 9 did indeed do many things right; in fact, many things are done in a very Unixy way. For example, the networking stack is build precisely where it ought to be: as a detail below a powerful network protocol (9P) which can be made as transparent or concrete as necessary (one can easily write individual TCP or UDP packets using the file system; just as easily, you can network-export any file system).

Similarly, Plan 9's handling of security, in particular the elimination of the superuser, was fantastic. And while I in particular disagree with the approach Plan 9 takes to the GUI, it is certainly a consistent, powerful, and inspired model. And a variety of its other ideas, such as UTF-8 and the /proc and /sys file systems, have worked their way into modern Unixes such as Linux.

Now, Plan 9's failure has been aptly explained elsewhere, in particular in the article linked. But, being somewhat old (2004, I believe; though when that particular section was written I do not know), it makes the prediction that Unixes will eventually come to absorb all of the Plan 9 features. Instead, it looks like the Linux kernel, at least, has mostly given up on ever implementing union mounts (one of the most inspired of Plan 9's innovations, at least in my view). And while the /net file system is incredibly useful, its power is much diminished when you can't assume that the world speaks 9P, which will seemingly never be the case. And a variety of Plan 9's choices are impossible without ignoring POSIX compatibility, which causes problems of its own.

Unix rode the growth of servers in the seventies and eighties, and is by now rather entrenched (see the History chapter in Art of Unix Programming); much like the WIMP GUI metaphor rode the personal computing wave. I think both of these areas are by now too old and entrenched for us to simply replace the current monopolies there. And I've come to terms with the fact that Plan 9 will never, ever become the standard OS --- it is simply not designed for any battlegrounds except those it has already lost in. But I strongly urge anyone designing an OS (or, really, any sort of all-encompassing computer system) to look over the innovations and ideas of Plan 9. Luckily, I don't think anything like Windows or Unix will win in a totally new environment, because Windows and Unix aren't made for any environments but their own. So while Plan 9 never won, I think its ideas have a good chance of living on. And I await an exciting time of innovation in Operating Systems in the coming century.

cturner• 14 years ago

    > And I've come to terms with the fact that Plan 9 will
    > never, ever become the standard OS --- it is simply
    > not designed for any battlegrounds except those it
    > has already lost in

Why is 9p cool? Are there any gaps - for example - lack of good support from scripting languages?

Have there been any serious attempts at creating a plan 9 rackspace or cloud service? Maybe you could get a hobbyist community grown around plan 9 + 9p.

I haven't looked at plan9 for ages, but remember there was some complexity about setting it up for ssh access.

Update: fgudin pointed me to this for hosting http://sdf.org/?join

ZephyrP• 14 years ago

I'm still rocking Plan9.

jff5• 14 years ago

9P is cool, but 9P is not Plan 9. 9P is the Plan 9 filesystem protocol.

Plan 9 lacks emacs, bash, C++, and a full-featured web browser; this alone is enough to make a lot of people post once on the mailing list bitching about it, then never come back. It does have vim now, although its use is not encouraged because we have other editors.

If you come at Plan 9 as though it's "just like Unix", you're going to have a bad time. Expecting to access it via ssh is one part of that--yes, we have an old ssh server, and yes we now have ssh v2 client support, but to access a Plan 9 system you want to use something like drawterm on Linux/Windows/Mac or cpu from another Plan 9 system.

I'm on sdf but have not played with my Plan 9 instance much. If you want to start off experimenting, I'd recommend just firing up VMware or Virtualbox instead. If you decide you want a physical Plan 9 system, well, I've found that it runs pretty well on every Thinkpad I've ever tried.

If you want more info, you can ask here or email me (check my profile).

asolove• 14 years ago

Where Plan 9 (and Inferno) live on most visibly is in the non-OS space:

- FUSE for file systems that are generated programmatically or exportable over a network. [http://code.google.com/p/macfuse/]

- The approach to concurrency and network programming in the Go language. [http://golang.org/]

Both of which show that problems that we used to leave to the OS can be elegantly solved in userland, which is a good thing.

jbarham2• 14 years ago

Unsurprisingly given the common pedigree, a number of ideas from Plan 9 have made their way into Go. E.g., compare http://golang.org/pkg/net/#Dial and http://plan9.bell-labs.com/magic/man2html/2/dial.

Go's approach to concurrency is familiar in the Plan 9 world from Alef (http://en.wikipedia.org/wiki/Alef_(programming_language), http://swtch.com/~rsc/thread/alef.pdf) and Limbo (http://www.vitanuova.com/inferno/papers/limbo.html), the language used to implement Inferno, a Plan 9 spinoff.

dhconnelly• 14 years ago

Cool. I wasn't sure if it was common knowledge that Ken Thompson and Rob Pike created Go. The Go compiler toolchain is from Plan 9, too: http://golang.org/cmd/.

thristian• 14 years ago

The impression I got was that the guys behind Unix were annoyed when their operating-systems research was stalled in the name of "API compatibility", so they started Plan 9 and actively resisted commercialisation for as long as possible (for example, they gave it the least marketable name they could think of).

There's so many interesting ideas in Plan 9, I'm occasionally tempted to install it and try it out... but then I read things like the list of keyboard shortcuts in the standard editor acme¹, and I get cold feet.

¹: http://acme.cat-v.org/keyboard - in particular, note that there are no keyboard shortcuts for 'move the cursor to the next/previous line'. Think about that.

4ad3• 14 years ago

The problem wasn't that the Plan 9 guys opposed commercialization, quite the contrary, the problem was that AT&T never let the system become open source. For some time it wasn't even available outside universities. That killed it.

Yes, acme doesn't have keyboard shortcuts, I use it every day.

hollerith2• 14 years ago

>the problem was that AT&T never let the system become open source.

The whole thing except the fonts was put under a very liberal open-source licence 10 or 12 or so years ago. (By then it was owned by Bellcore or Lucent, not AT&T.) Before then there were severe restrictions; is your opinion that by time the severe restrictions were lifted, it was too late because Linux had already become entrenched?

jff5• 14 years ago

Regarding acme, everyone comes in and goes "Yuck, you mean I can't type 'C-c M-x undoify'?" However, once you get used to it, it's actually rather nice. I have something like 30-40 files open in acme right now, and I find it far, far easier to manage open buffers than in Emacs. Arranging and re-arranging buffers (to use the emacs term, since it's familiar) is very convenient and helps me keep things organized.

It also has "mouse shortcuts" (chords) that tend to reward using the mouse more--meaning you don't need to switch from mouse to keyboard as often, because you can do cut, copy, paste, search, command execution, etc. all with a click of the mouse (and no hunting through menus, either).

It's hard to explain... you really have to use Acme, or even better, watch someone use it. http://www.youtube.com/watch?v=dopu3ZtdCsg seems to be a decent intro, although I haven't had time to watch the whole thing yet.

gnosis• 14 years ago

C-c M-x undoify ?

In vim it's just: u

Much simpler and quicker than stretching your arm to the mouse, futzing around with the cursor to find god knows what menu option, clicking on it, and moving your hand back to the keyboard.

Having 30-40 files open in an editor is not impressive either. You can easily have hundreds of files open in vim, and navigate between them without a problem.

The mouse chords might make the mouse more functional, but it's certainly a far cry from having over 100 keys and thousands of key combinations at your fingertips.

Not that using the mouse excludes the possibility of using the keyboard, but as a vim user I almost never have need to use the mouse -- except when I've switched to vim from another application that requires mouse use and want to paste something. After that, my hand goes right back to the keyboard, because using the keyboard is just far more efficient.

Acme's forcing mouse use on the user is one reason I am really not very interested in using it. However, I am interested in seeing if there is anything it can do that doesn't absolutely require mouse use that could be integrated in to vim.

bitwize2• 14 years ago

If AT&T could successfully market an OS with a name that sounds like "eunuchs", they weren't going to be dissuaded by something innocuous like Plan 9.

And indeed, Plan 9 is conceptually cool as balls, but Acme and Sam make me want to chuck expensive electronic equipment out the window.

gioele• 14 years ago

Plan 9 was and is too good to be ignored. Indeed many of its features are now a must in Linux systems (for example good support for private namespaces) and the p9fs, designed to be virtualizable and distributed, is getting serious traction in the virtualization world (see 9p virtio).

And Plan 9 also has the best mascot around!

mdwrigh2• 14 years ago

It helps that Eric Van Hensbergen, the guy behind 9p virtio, was a Plan 9 developer.

4ad3• 14 years ago

s/was/is/

He was one of the guys that ported Plan 9 to the Blue Gene supercomputer. He has many active projects in the Plan 9 community.

rbanffy5• 14 years ago

This need updating. "The Room", a film that came after the article was written, is obviously worse than either "Manos" or "Plan 9", as anyone who saw even its first minutes will readily attest.

edit: I'm sorry. I often forget my colleagues are humourless at this time in the morning.

astrange• 14 years ago

I think "Birdemic" would be more appropriate, since the first half is about someone pitching a tech startup (and the second half is about being vomited on by animated GIFs of birds).

rbanffy5• 14 years ago

I can't disagree with facts. According to IMDB, http://www.imdb.com/title/tt1316037/ is worse than http://www.imdb.com/title/tt0368226/

haberman2• 14 years ago

Am I the only one who doesn't get excited about the idea of absolutely everything being a file?

People seem to like the idea of being able to say things like "cat file.wav > /dev/dsp". But a sound card is a complex thing that can be configured in umpteen different ways; what if the file has a different sample rate than the card is currently running at? Even if you gerry-rigged a kernel driver that could handle this case (by, for example, noticing when the file had been open()'d and expecting the next bytes to be a WAV header, which it uses to reconfigure the card), you're left with something that doesn't have even the most basic functionality that you'd expect from a sound player; for example, "pause."

So what has representing /dev/dsp as a file really bought you? It doesn't help you know what devices are available, because there's tons of crap in /dev that doesn't necessarily have a kernel driver loaded for it. It's not that useful as an end-user interface for selecting a sound card, because users don't want to choose between /dev/dsp and /dev/dsp2, they want to see nice names like "Internal Speaker" or "Apogee Duet."

Files give you a simple API for lots of common operations, which is nice, but the fact that all files are not created equal means that operations sometimes don't work like you expect. For example, how would you guess that "tail -f" is implemented? If you're like me, you would have expected that it works by calling poll/select/etc. and waiting for the file to become read-ready. But this approach doesn't actually work for monitoring growing files; the select() will just return immediately with read-ready even if you're at EOF. So what "tail -f" actually does is a loop that alternates between sleep() and fstat(). I'm serious! http://git.savannah.gnu.org/gitweb/?p=coreutils.git;a=blob;f...

strictfp2• 14 years ago

You could configure it by doing something like

  echo "22050" > /dev/dsp/config/sample_rate_hz
  echo "lanczos" > /dev/dsp/config/resample_algorithm

You could then search for configs by using find, back them up by using tar, be able to easily replicate settings between computers using rsync etc.

The idea is not to have one unconfigurable binary reciever, but rather to expose the interface as a hierarchy of rw files instead of a number of C functions, because the C functions have no easy shell representation, and most command-line tools fail using them.

It's very reminiscent of REST.

adobriyan4• 14 years ago

> The idea is not to have one unconfigurable binary reciever,

It _is_ configurable, why do you think it is not?

This idea that somehow one can fit any interaction with hardware into a filesystem ends when hardware becomes complex enough to simply NOT fit (like modern GPU).

How do you guarantee atomicity when saving configuration with tar(1)?

Linux recently deprecated textual conntracking data interface in favour of shipping binary data over netlink socket. Reason -- read() speed in real life high load scenarios.

What could Plan9 do here?

dalke2• 14 years ago

That API seems so clunky. What if I have two different programs, with different sample rates, which I want to play back at the same time?

pmarin• 14 years ago

http://plan9.bell-labs.com/magic/man2html/3/audio

djnliung2• 14 years ago

I'm not sure that really contradicts his point, that's a terribly limited API. Stereo only, no way to specify channel layout, 16bit only, no way to change the device buffer sizes, no way to tell if the device is in use, no way to tell the device latency, seemingly no way to examine what sample rates or bit depths the device supports, seemingly no way to subscribe to any kind of notification when the device has actually changed sample rate (or other properties) or when playback has stopped etc, no ability to lock the device to prevent other processes changing the sample rate or other properties, no ability to get device time or determine if the device is synchronised with system time.

Some of these missing things can be added fairly easily to the existing API of course, but I don't know enough about plan9 to suggest how, for example, notifications for device property changes could be added.

As it stands this interface is almost useless even for casual home users who just want to play back audio: multiple channel support is a basic requirement these days. Serious audio work is not even possible at all.

beagle3• 14 years ago

Some files are different. So? Do you have the same qualms with stdin/stdout not being seekable because they are often connected to a terminal?

The "everything is a file" abstraction is mostly helpful and rarely gets in the way in practice; I don't know how plan9 exposes sound, but it is probably a directory, with multiple files - one you write to to set sampling rate; another gets data, and the third is stop/go/cancel.

The most useful feature, though, is network transparency that really works. You don't need Vnc/rdp; just mount the remote display in your file system. You don't need special ip forwarding - just mount the remote network in your local file system. You don't need a network aware server for anything - it all is.

That's the reason to get excited.

vidarh• 14 years ago

You can do cat file.au >/dev/audio on a lot of Unix-y OS's. For wav, the problems you mention are solved with "sox" which converts the wav file to raw data with suitable options.

All of the problems you mention are problems of the interfaces exposed by the files, not problems with the "everything as a file" idea in itself.

Look at /proc on linux. /proc/[pid] represents a running process, but it's a directory, not a file, all the information about that process are inside.

A hypothetical filesystem API for audio designed to be friendly might expose a /proc/sound/[device]/name file containing a user friendly description and a /proc/sound/[device]/output file that a converter gets mounted at that you can just copy files to, and a /proc/sound/[device]/control that accepts commands like "pause" or "play".

You also seem to also be assuming that these interfaces are meant for ordinary end-users. While some of the functionality does make it easier for end-users to interact directly with the lower levels of the system, nothing precludes friendlier interfaces being put on top of it.

What it gives you is an interface that has a bunch of standard tools you can use to operate on them, either because they're simple enough to use directly, or to make building the user friendly interfaces more easily, and that which is at the same time reasonably explorable if it's well designed. /proc is a good example of this - you can poke around in it and learn a lot about your system with just a standard shell without having the faintest idea about how it's structured initially. Meanwhile, the non-file-related system calls require you to write code to test them out - they're far more opaque.

> But this approach doesn't actually work for monitoring growing files; the select() will just return immediately with read-ready even if you're at EOF.

This really shouldn't be surprising. select() specifically tells you if the filedescriptor is read for read. A filedescriptor is ready to read from if there is more data or you're at the end of the stream. The behavior is entirely consistent. Perhaps a way of requesting notification only if there is more data to read, rather than if there's more data or end of the stream, would be useful, but that is not what select() provides. It illustrates some of the flaws in the Unix system calls, not an inherent problem with a filesystem interface.

wslh• 14 years ago

> But a sound card is a complex thing that can be configured in umpteen different ways; what if the file has a different sample rate than the card is currently running at?

That's what ioctl does: http://www.manpagez.com/man/2/ioctl/

bluetech• 14 years ago

Well, just below the Plan 9 sections comes this: http://catb.org/~esr/writings/taoup/html/ch20s03.html#id3016... And he's right, ioctl's can be quite tricky to operate sometimes.

haberman2• 14 years ago

Sure, you can use an ioctl for this. You could use some combination of ioctl()s to tunnel absolutely any API, just like you can wrap any text string in pointy brackets and call it XML.

The question is whether this is useful. In particular, once your interface requires ioctl(), you lose the benefit of easy interoperability with UNIX command-line tools.

clord• 14 years ago

The advantage of "everything-is-a-file" is that every process gains thread-safe access to some API without having to pull in the headers, link, and manage a foreign memory management regimen.

But the real boon is that each process can conceivably be linked with any 'file'. I liken this practice to the STL library's iterator pattern: algorithms are the processes, and files are the iterators (although I bet STL was inspired by unix). You can redirect your tty through an ssh tunnel, you can scp a regular file, or you can ssh tunnel your soundcard or mouse. And of course 'ssh tunnel' is only one such kind of glue. There are pipes, devices, fifos, etc. In the end it works out that there are O(n*m) possibilities for only O(n+m) implementation cost.

But remember that with STL, not all algorithms will work with all iterators, and vice versa. And likewise, not all processes can work with all files, and vice versa. The benefit comes from the significant overlapping portion.

ori_b3• 14 years ago

>Am I the only one who doesn't get excited about the idea of absolutely everything being a file?

Don't think of it as everything being a file. Think of it as accessing everything including file data through an RPC system.

andrewflnr• 14 years ago

It's been a while since I looked at 9P. Is it really that much like an RPC system?

jdthomas• 14 years ago

The HUGE advantage of "everything is a file" is that if you make a tool that works on one "thing" and then want to use it on another "thing" it is quit possible that it will just work. For instance, on OSX, one of the things I sorely miss is the sysfs from Linux. Sure, OSX has the ioreg, but I cannot use the tools I am used to (find, grep, cat, echo, sed, awk, ...).

To put it in terms the web programmers can understand, it is like REST, you have a GET (read), and a POST (write), and I suppose a DELETE (unlink), and with that you can implement all you need. Everything fits into a very few syscalls (open, read, write, close).

gosub• 14 years ago

    > There is a lesson here for ambitious system architects: the 
    > most dangerous enemy of a better solution is an existing
    > codebase that is just good enough.

Isn't this concept applicable to every system once it incorporates programmability?

mhw• 14 years ago

I've often thought that Plan 9's everything-is-a-file system with a consistent API for access is the missing link between Unix's everything-is-a-file and REST's consistent interface constraint. I think it's certainly arguable that Plan 9 shares many features with RESTful architectures.

shimon_e2• 14 years ago

The real shame is that inferno never became android. It was around early enough. Had the right feature set. I guess it was not in Google's backyard (America) and had lost it's backbone a little to early.

I guess when Google was shopping for mobile OS they were also looking for a talent grab. With Inferno by 2005 most had already left.

sambeau• 14 years ago

True, but, they already had all the talent from Plan 9

luriel• 14 years ago

By 2005 most of the team that created Plan 9 and Inferno at Bell Labs were already working at Google.

But they were working on other stuff, like Go.

mhd3• 14 years ago

I always find it pretty sad that while everybody speaks about the integrated environments of Smalltalk and Plan 9, Niklaus Wirth's Oberon is almost left forgotten…

hollerith2• 14 years ago

For over a decade, dozens of actual people used Plan 9 as their main environment. (Many but not all of these people worked at Bell Labs.) I tend to believe that Smalltalk has also seen serious numbers of actual users. In contrast, who has ever used Oberon as their main software environment except perhaps students required to do so to pass some class?

mhd3• 14 years ago

If we're talking about popularity contests here, all the "contestants" aren't doing very well…

adbge• 14 years ago

One of the Plan 9 papers actually refers to Oberon as an example of a similarly integrated system. I believe it's the one titled 'The Use of Name Spaces in Plan 9,' but I'm not certain.

Unfortunately, the server is down right now so I'm unable to check but, if you're curious, the various papers can be found here: http://plan9.bell-labs.com/sys/doc/

mhd3• 14 years ago

Probably the acme paper. Oberon definitely had the mouse chording, not sure if there's previous art in that regard.

Craiggybear2• 14 years ago

"Those who don't understand UNIX are condemned to reinvent it. Poorly." – Henry Spencer

atuladhar• 14 years ago

Except that Plan 9 was designed by the people who built UNIX. Are you implying that the people who built UNIX don't understand it?

Craiggybear2• 14 years ago

No, no! Quite the reverse.

runjake• 14 years ago

Plan 9 wasn't a failure and it wasn't the future, directly.

It was a research OS that was never even close to being positioned to market, be it consumers, servers, etc -- although technology coming out of Plan 9 did make it into some commercial applications.

Indirectly, Plan 9 technologies (or inspired successors) made their way into real products. Linux's procfs is one notable example.

morphles• 14 years ago

I recently tried plan 9 for a bit. Though i understand all the cool things it does and agree with a lot of it. One thing really disappointed and struck me as totally brain dead: there is no such thing as terminal emulator or it seems something like text only mode. In general it seems to me that achieving keyboard only control (that is comfortable to use) is not really possible at least for now.

angli• 14 years ago

A good introduction to the Plan 9 UI I found, for any who are interested: http://vimeo.com/7748726

Plan 9: The way the future was

Related Stories