[9fans] p9sk2

[9fans] p9sk2

[Uriel]

I was trying to document -O in cpu(1) which is undocumented, it enables
'p9sk2', but what that does is rather confusing, my initial guess was that it
was a deprecated version of p9sk1,  but lookman p9sk2 pointed me at
factotum(4) which says:

p9sk1   a Plan 9 shared key protocol described in
      authsrv(6)'s ``File Service'' section.
p9sk2   a variant of p9sk1 described in authsrv(6)'s
      ``Remote Execution'' section.

(Oh, and it seems that factotum(4) in p9p and in Plan 9 are out of sync..
*sigh*)

Looking at authsrv(6) leaves me even more confused. 

But auth/factotum/p9sk1.c seems to confirm the original theory:

% grep p9sk2 /sys/src/cmd/auth/factotum/p9sk1.c|sed 1q
 * p9sk1, p9sk2 - Plan 9 secret (private) key authentication.

This all started because cpu -O is used in
http://plan9.bell-labs.com/wiki/plan9/Drawterm_to_your_terminal/

My guess is that it's needed for drawterm to be able to connect, will dt2k
fix that? what is up with dt2k anyway?

uriel

Re: [9fans] p9sk2

[Axel Belinfante]

> I was trying to document -O in cpu(1) which is undocumented,

> My guess is that it's needed for drawterm to be able to connect, will dt2k
> fix that? what is up with dt2k anyway?

I looked for -O and -R in the source yesterday,
when I was trying enable not only drawterm but also
cpu to the terminal, for which one needs listeners
on ports cpu and ncpu respectively,
which call cpu with -O and -R respectively.

what I remember is that -O starts old9p which
translates between the old 9p and 9p2000.

Axel.

Re: [9fans] p9sk2

[Axel Belinfante]

I wrote to quickly:
> > I was trying to document -O in cpu(1) which is undocumented,
> 
> > My guess is that it's needed for drawterm to be able to connect,
> > will dt2k fix that? what is up with dt2k anyway?
> 
> I looked for -O and -R in the source yesterday,
> when I was trying enable not only drawterm but also
> cpu to the terminal, for which one needs listeners
> on ports cpu and ncpu respectively,
> which call cpu with -O and -R respectively.
> 
> what I remember is that -O starts old9p which
> translates between the old 9p and 9p2000.

in addition it indeed seems to select p9sk2
instead of p9any as auth proto.
(I seem to recall that p9sk2 was discussed here
 some time ago).

Axel.

Re: [9fans] p9sk2

[Uriel]

On Tue, Oct 04, 2005 at 11:30:27PM +0200, Axel Belinfante wrote:
> > what I remember is that -O starts old9p which
> > translates between the old 9p and 9p2000.
> in addition it indeed seems to select p9sk2 instead of p9any as auth
> proto.
This much I had figured out...
 
> (I seem to recall that p9sk2 was discussed here some time ago).
Searching all I can find is
http://groups.google.com/group/comp.os.plan9/msg/b031d22ac909c3d7
which seems to indicate the man pages have been somewhat confused for
years.

uriel

Re: [9fans] p9sk2

[Russ Cox]

> I was trying to document -O in cpu(1) which is undocumented, it enables

It is like -R (run the server aka remote side of the protocol)
except that it runs the pre-9P2000 server side.
It is undocumented because you never say it yourself
on the command line (just like -R) and because it needs
to go away.  It persists only because of drawterm.

> 'p9sk2', but what that does is rather confusing, my initial guess was that it
> was a deprecated version of p9sk1,  but lookman p9sk2 pointed me at
> factotum(4) which says:
>
> p9sk1   a Plan 9 shared key protocol described in
>       authsrv(6)'s ``File Service'' section.
> p9sk2   a variant of p9sk1 described in authsrv(6)'s
>       ``Remote Execution'' section.

Authsrv(6) is still describing the pre-9P2000 protocols.
Except for the 9P stuff, it's still accurate and does describe
p9sk1 and p9sk2 (without naming them) in the sections
mentioned.

/sys/doc/auth.ps has a quick sketch of p9sk1 that may well
be easier to digest, and
http://swtch.com/usr/local/plan9/src/cmd/auth/factotum/p9sk1.c
has a similar sketch and tells you exactly what messages are
in play in both.

P9sk2 exists only for backwards compatibility when connecting
to old (pre-9P2000) services.  It does not authenticate the server
to the client, only vice versa.  It was used by cpu in the old days,
as you've no doubt figured out.

> (Oh, and it seems that factotum(4) in p9p and in Plan 9 are out of sync..
> *sigh*)

Both factotum(4)s accurately describe their respective factotums.
Go ahead.  Complain that the factotum programs are out of sync too.

> This all started because cpu -O is used in
> http://plan9.bell-labs.com/wiki/plan9/Drawterm_to_your_terminal/
>
> My guess is that it's needed for drawterm to be able to connect, will dt2k
> fix that? what is up with dt2k anyway?

As it always is, progress is slow.  Skip and Brucee have added Windows
support to dt2k, and Andrey has been maintaining it.  I use it for days
at a time without problems on Linux except that snarf doesn't work.
Others have reported problems.  At the moment, it's really far down
on my to do list, and no one else is stepping up to the plate.

Russ

Re: [9fans] p9sk2

[Skip Tavakkolian]

> what I remember is that -O starts old9p which
> translates between the old 9p and 9p2000.

cpu -O causes cpu to talk to drawterm with srvold9p betwixt.
i drew a .pic of it once, if you want it.

Re: [9fans] p9sk2

[andrey mirtchovski]

> Skip and Brucee have added Windows
> support to dt2k, and Andrey has been maintaining it.  I use it for days
> at a time without problems on Linux except that snarf doesn't work.
> Others have reported problems.  At the moment, it's really far down
> on my to do list, and no one else is stepping up to the plate.

i'm also using dt2k without any trouble (even copy/paste seems to work
for me just fine, in linux).  i've switched to it exclusively and
haven't had it crash since the last recompile which happened a month
or two ago.

andrey

Re: [9fans] p9sk2

[geoff]

Is the drawterm.tgz on your web page up to date (and is it dt2k)?
Should I generate a new round of binaries?

Re: [9fans] p9sk2

[Uriel]

On Tue, Oct 04, 2005 at 05:44:04PM -0400, Russ Cox wrote:
> > I was trying to document -O in cpu(1) which is undocumented, it enables
> 
> It is like -R (run the server aka remote side of the protocol)
> except that it runs the pre-9P2000 server side.
> It is undocumented because you never say it yourself
> on the command line (just like -R) 
I'm not sure that just because you don't call it directly it means that
it should not be documented, you still need to use some times, see:
http://plan9.bell-labs.com/wiki/plan9/Drawterm_to_your_terminal/

> and because it needs  to go away. It persists only because of
> drawterm.
On that I will agree.

> > My guess is that it's needed for drawterm to be able to connect, will dt2k
> > fix that? what is up with dt2k anyway?
> 
> As it always is, progress is slow.  Skip and Brucee have added Windows
> support to dt2k, and Andrey has been maintaining it.  I use it for days
> at a time without problems on Linux except that snarf doesn't work.
> Others have reported problems.  At the moment, it's really far down
> on my to do list, and no one else is stepping up to the plate.
First one would have to find the source... wasn't drawterm(2k) supposed
to move to sources?

I know f2f had the source in his site, but it's not linked from
anywhere, and I can't recall the address, and that was a while ago, so
who knows if the code is up to date?

The stuff in /sys/src/cmd/unix/drawterm looks quite outdated.

uriel

Re: [9fans] p9sk2

[andrey mirtchovski]

> Is the drawterm.tgz on your web page up to date (and is it dt2k)?
> Should I generate a new round of binaries?

no, those are still old-9p drawterm binaries.  russ has set up a cvs
repository for dt2k which i commit my changes to, if i have any.  i
don't know whether he has made it available anywhere for download.

Re: [9fans] p9sk2

[Skip Tavakkolian]

> i'm also using dt2k without any trouble (even copy/paste seems to work
> for me just fine, in linux).  i've switched to it exclusively and
> haven't had it crash since the last recompile which happened a month
> or two ago.

i've use dt2k on windows all the time.  the only problem i know of is
the keyboard alt+char input.  it isn't fatal as i can cut and paste
out of /lib/keyboard.  i think somebody already fixed this. 

Re: [9fans] p9sk2

[Russ Cox]

> > i'm also using dt2k without any trouble (even copy/paste seems to work
> > for me just fine, in linux).  i've switched to it exclusively and
> > haven't had it crash since the last recompile which happened a month
> > or two ago.
>
> i've use dt2k on windows all the time.  the only problem i know of is
> the keyboard alt+char input.  it isn't fatal as i can cut and paste
> out of /lib/keyboard.  i think somebody already fixed this.

this has to be an issue of just not calling the latin1 code
from the windows keyboard driver.  it works on linux.

russ

Re: [9fans] p9sk2

[andrey mirtchovski]

i have no idea how divergent the X code is from the windows one  as
far as functionality is concerned -- i've added things from the old
drawterm such as keysym input and scroll-button support.

andrey

Re: [9fans] p9sk2

[Skip Tavakkolian]

> As it always is, progress is slow.  Skip and Brucee have added Windows
> support to dt2k, and Andrey has been maintaining it.  I use it for days
> at a time without problems on Linux except that snarf doesn't work.
> Others have reported problems.  At the moment, it's really far down
> on my to do list, and no one else is stepping up to the plate.

i think richard added the ssl to dt2k.

also, brucee -- as part of 9netics work -- added aan support to a dt2k
derivative.  if geoff, andrey (or russ) want to do it, it would be a
clean operation to take the changes and patch dt2k.  there is a new
aanfs and cpu.c that goes with it.  it is experimental. it is not fully tested.

Re: [9fans] p9sk2

[Russ Cox]

> > As it always is, progress is slow.  Skip and Brucee have added Windows
> > support to dt2k, and Andrey has been maintaining it.  I use it for days
> > at a time without problems on Linux except that snarf doesn't work.
> > Others have reported problems.  At the moment, it's really far down
> > on my to do list, and no one else is stepping up to the plate.
>
> i think richard added the ssl to dt2k.
>
> also, brucee -- as part of 9netics work -- added aan support to a dt2k
> derivative.  if geoff, andrey (or russ) want to do it, it would be a
> clean operation to take the changes and patch dt2k.  there is a new
> aanfs and cpu.c that goes with it.  it is experimental. it is not fully tested.

you just made my day.

russ

Re: [9fans] p9sk2

[Russ Cox]

> > As it always is, progress is slow.  Skip and Brucee have added Windows
> > support to dt2k, and Andrey has been maintaining it.  I use it for days
> > at a time without problems on Linux except that snarf doesn't work.
> > Others have reported problems.  At the moment, it's really far down
> > on my to do list, and no one else is stepping up to the plate.
>
> i think richard added the ssl to dt2k.

he added ssl to the 9P1 drawterm.
it was always part of dt2k, since in 9P2000 cpu,
encryption is required.  (in old cpu it is merely optional.)

the cvs repository that andrey mentioned is at
http://cvs.pdos.csail.mit.edu/cvs/drawterm

the checkout procedure is the same as plan 9 from user space,
replacing "plan9" with "drawterm" in the cvs command.

russ

Re: [9fans] p9sk2

[David Leimbach]

On 10/4/05, Skip Tavakkolian <9nut@9netics.com> wrote:
> > As it always is, progress is slow.  Skip and Brucee have added Windows
> > support to dt2k, and Andrey has been maintaining it.  I use it for days
> > at a time without problems on Linux except that snarf doesn't work.
> > Others have reported problems.  At the moment, it's really far down
> > on my to do list, and no one else is stepping up to the plate.
>
> i think richard added the ssl to dt2k.
>
> also, brucee -- as part of 9netics work -- added aan support to a dt2k
> derivative.  if geoff, andrey (or russ) want to do it, it would be a
> clean operation to take the changes and patch dt2k.  there is a new
> aanfs and cpu.c that goes with it.  it is experimental. it is not fully tested.
>

aan support is, if memory serves, the connection stability stuff.  If
so, that's really super-excellent.  Awesome!  Happy happy joy joy!

[cutting back on my caffeine intake now...]

Dave

Re: [9fans] p9sk2

[Uriel]

On Tue, Oct 04, 2005 at 07:35:59PM -0400, Russ Cox wrote:
> he added ssl to the 9P1 drawterm.
> it was always part of dt2k, since in 9P2000 cpu,
> encryption is required.  (in old cpu it is merely optional.)
> 
> the cvs repository that andrey mentioned is at
> http://cvs.pdos.csail.mit.edu/cvs/drawterm
I thought I had looked there and could not find it, I need to have my
eyes checked...

Can we kill /n/sources/plan9/sys/src/cmd/unix/drawterm? it's outdated
and misleading. 

While we are at it,9pfreebsd provably should move to sources/extra/old/
or something like that as it's only of historical interest and there is
no point in wasting space in the default source tree; and tcs.shar.Z can
probably die as it's part of p9p this days.

If someone could put a set of precompiled drawterm binaries in sources
it would be great, the least sites we have to keep track of, the better.

I went and created a stub for a wiki drawterm page:
http://plan9.bell-labs.com/wiki/plan9/drawterm/

uriel

Re: [9fans] p9sk2

[Russ Cox]

> Can we kill /n/sources/plan9/sys/src/cmd/unix/drawterm? it's outdated
> and misleading.

Outdated, yes.  Misleading, no.  Once dt2k is a suitable
replacement, I will put it there.  Dt2k is still a work in progress
and the old drawterm still works.

> If someone could put a set of precompiled drawterm binaries in sources
> it would be great, the least sites we have to keep track of, the better.

Eventually this will happen.  Things move slowly, but with
such little time to do any work, there's no point in doing
work before it's time.  Of course, if you're volunteering, ...

Russ

Re: [9fans] p9sk2

[Skip Tavakkolian]

> but with
> such little time to do any work, there's no point in doing
> work before it's time.

this seems good enough to be a fortune.

Re: [9fans] p9sk2

[Uriel]

On Tue, Oct 04, 2005 at 08:19:21PM -0400, Russ Cox wrote:
> > Can we kill /n/sources/plan9/sys/src/cmd/unix/drawterm? it's outdated
> > and misleading.
> 
> Outdated, yes.  Misleading, no.  Once dt2k is a suitable replacement,
> I will put it there.  Dt2k is still a work in progress and the old
> drawterm still works.
As others have noted dt2k seems to work fine, I just checked it, not
only it seems to work fine and feels much faster than the old drawterm,
it fixes most of the really annoying bugs that plagued the old drawterm,
it also simplifies things because we can forget about the old 9P.

If dt2k is not perfect yet, I don't care, it's much better than the old
drawterm, of course hiding it in a vault I doubt anyone will test it.

> > If someone could put a set of precompiled drawterm binaries in sources
> > it would be great, the least sites we have to keep track of, the better.
> Eventually this will happen.  Things move slowly, but with
> such little time to do any work, there's no point in doing
> work before it's time.  Of course, if you're volunteering, ...
I have put Linux and FreeBSD binaries in:
/n/sources/contrib/uriel/drawterm/2005-10-05/

I will put the OpenBSD binaries tomorrow once I install X on my obsd
box. If someone has other binaries(specially windows) please send them
to me and I will put them there.

Release early, release often, and if you don't like it, look at Google.

uriel

Re: [9fans] p9sk2

[Russ Cox]

> As others have noted dt2k seems to work fine, I just checked it, not
> only it seems to work fine and feels much faster than the old drawterm,
> it fixes most of the really annoying bugs that plagued the old drawterm,
> it also simplifies things because we can forget about the old 9P.

Seems to work fine and actually works fine are two different things.

> If dt2k is not perfect yet, I don't care, it's much better than the old
> drawterm, of course hiding it in a vault I doubt anyone will test it.

You may not care, but I do.  And web links to dt2k have been in
a bunch of 9fans postings, perhaps from before your time but
that's what search engines are for.

> I have put Linux and FreeBSD binaries in:
> /n/sources/contrib/uriel/drawterm/2005-10-05/
> I will put the OpenBSD binaries tomorrow once I install X on my obsd
> box. If someone has other binaries(specially windows) please send them
> to me and I will put them there.

Thank you.

Russ

Re: [9fans] p9sk2

[jmk]

	>Release early, release often, and if you don't like it, look at Google.

wtf does that mean?

Re: [9fans] p9sk2

[LiteStar numnums]

[-- Attachment #1: Type: text/plain, Size: 936 bytes --]

Just to gander a guess, but possibly the whole 'everything is beta forever'
wonder that is google?

On 10/4/05, jmk@plan9.bell-labs.com <jmk@plan9.bell-labs.com> wrote:
>
> >Release early, release often, and if you don't like it, look at Google.
>
> wtf does that mean?
>



--
The subject of this essay (the Myth of Sisyphus) is precisely
this relationship between the absurd and suicide, the exact
degree to which suicide is a solution to the absurd. The
principle can be established that for a man who does not cheat,
what he believes to be true must determine his action.
Belief in the absurdity of existence must then dictate his
conduct. It is legitimate to wonder, clearly and without
false pathos, whether a conclusion of this importance
requires forsaking as rapidly possiblean imcompre-
hensible condition. I am speaking, of course, of men
inclined to be in harmony with themselves.
<< Albert Camus>>

[-- Attachment #2: Type: text/html, Size: 1375 bytes --]

Re: [9fans] p9sk2

[Nils O. Selåsdal]

On Wed, 2005-10-05 at 03:45 +0200, Uriel wrote:
> On Tue, Oct 04, 2005 at 08:19:21PM -0400, Russ Cox wrote:
> > > Can we kill /n/sources/plan9/sys/src/cmd/unix/drawterm? it's outdated
> > > and misleading.
> > 
> > Outdated, yes.  Misleading, no.  Once dt2k is a suitable replacement,
> > I will put it there.  Dt2k is still a work in progress and the old
> > drawterm still works.
> As others have noted dt2k seems to work fine, I just checked it, not
The last I tried of it (3-4 months ago), it worked fine until it
segfaults ... usually within an hour.

-- 
Nils O. Selåsdal <NOS@Utel.no>

Re: [9fans] p9sk2

[Richard Miller]

> Others have reported problems.  At the moment, it's really far down
> on my to do list, and no one else is stepping up to the plate.

Where are the problems being reported to?  It may be more likely for
others to "step up" if they know that something needs doing (and that
nobody else is already working on it).

-- Richard

Re: [9fans] p9sk2

[Skip Tavakkolian]

> Where are the problems being reported to? 

9trouble, i would think.

Re: [9fans] p9sk2

[Russ Cox]

> > Others have reported problems.  At the moment, it's really far down
> > on my to do list, and no one else is stepping up to the plate.
>
> Where are the problems being reported to?  It may be more likely for
> others to "step up" if they know that something needs doing (and that
> nobody else is already working on it).

It's not like the bug reports are just flowing in: a crash here,
a crash there.  Mostly it was crashing while I used it, so I wasn't
actively soliciting crashes from others.  I used to get "unexpected
async message" from X every once in a while and then it would
hang.  Perhaps the latest version (from Andrey via the 9netics
crowd) has fixed these problems.  Andrey and I decided to stay
in sync via CVS instead of passing tar balls around and fix these
problems, but since we started on that I haven't seen it crash
once.  It does hang occasionally, but it comes back, so that's
probably the network.

I still don't have great confidence that it's okay.  As much as Uriel
wants to malign the old drawterm, I believe it was more solid than
this.

Russ

Re: [9fans] p9sk2

[Uriel]

On Tue, Oct 04, 2005 at 10:13:46PM -0400, jmk@plan9.bell-labs.com wrote:
>>Release early, release often, and if you don't like it, look at Google.
> 
> wtf does that mean?

"Design and build software, even operating systems, to be tried early,
ideally within weeks. Don't hesitate to throw away the clumsy parts and
rebuild them."

-- The Bell System Technical Journal. Bell Laboratories. M. D. McIlroy,
E.  N. Pinson, and B. A. Tague. "Unix Time-Sharing System Forward" 1978.
57 (6, part 2). p. 1902.

The google reference can be seen in recent interviews with Rob and
others about how their release process works.

I have heard of people complaining that working alone one doesn't get
much feedback... but what do I know, I'm just the buffoon of the court.

uriel

Re: [9fans] p9sk2

[jmk]

it was the 'look at google' reference i didn't get.

you're right about everything else.

On Mon Oct 24 09:53:11 EDT 2005, uriell@binarydream.org wrote:
> On Tue, Oct 04, 2005 at 10:13:46PM -0400, jmk@plan9.bell-labs.com wrote:
> >>Release early, release often, and if you don't like it, look at Google.
> > 
> > wtf does that mean?
> 
> "Design and build software, even operating systems, to be tried early,
> ideally within weeks. Don't hesitate to throw away the clumsy parts and
> rebuild them."
> 
> -- The Bell System Technical Journal. Bell Laboratories. M. D. McIlroy,
> E.  N. Pinson, and B. A. Tague. "Unix Time-Sharing System Forward" 1978.
> 57 (6, part 2). p. 1902.
> 
> The google reference can be seen in recent interviews with Rob and
> others about how their release process works.
> 
> I have heard of people complaining that working alone one doesn't get
> much feedback... but what do I know, I'm just the buffoon of the court.
> 
> uriel