Re: [9fans] python and lua

Re: [9fans] python and lua

[okamoto]

>we see problems even with a real file server, and suspect
>a bug in the paging code but haven't yet found it

What kind of situations?
I asked this because I've never seen it.

Kenji

Re: [9fans] python and lua

[okamoto]

[-- Attachment #1: Type: text/plain, Size: 398 bytes --]

>Try running a few instances of the following program
>
>	for(i in `{seq 10})
>		memeat -x 10000000

I tried this on our dual pentium(133MHz) without local disk jus now.

swap area (/n/gabbroother/swap/trachyte) grew up to 40MB, and I
stopped the program at that point.  Nothing bad happened.
I'll not try it on my notebook with local disk, because I know what
happenes.  :-)

Kenji


[-- Attachment #2: Type: message/rfc822, Size: 2769 bytes --]

From: "Russ Cox" <rsc@plan9.bell-labs.com>
To: 9fans@cse.psu.edu
Subject: Re: [9fans] python and lua
Date: Sun, 7 Jan 2001 23:12:50 -0500
Message-ID: <20010108041308.3BCC3199D7@mail.cse.psu.edu>

> What kind of situations?
> I asked this because I've never seen it.

Usually a terminal booted from local disk
can't sustain more than a few minutes of local disk swapping
before the window system or kfs swaps in a bogus page
and dies.  

Try running a few instances of the following program

	for(i in `{seq 10})
		memeat -x 10000000

and sit back.  This doesn't seem to happen when swapping
to a network file server.  Perhaps the timing is different.

I suspect something in the x86 memory management code,
although presotto says the carreras had similar problems.
I spent a day this summer trying to find anything, but
like everyone else came up empty handed.  

Instead, I moved more memory in from another machine
and turned off swapping.  On a laptop, of course,
this is less of an option, but I don't run my laptop
with swapping enabled either.  I'd much rather see 
"no physical memory" and type ^t^t k.

Russ


g% cat memeat.c
#include <u.h>
#include <libc.h>

void
main(int argc, char **argv)
{
	int xflag=0;
	int i, n;
	char *p;

	ARGBEGIN{
	case 'x':
		xflag=1;
	}ARGEND;
	n = atoi(argv[0]);
	p = malloc(n);
	memset(p, 0, n);
	i=0;
	for(;;) {
		sleep(1000+1000*(rand()%8));
		if(xflag)	memset(p, i++, n);
	}
}
g%

Re: [9fans] python and lua

[okamoto]

Thanks Russ of your detailed description.

>I'd much rather see 
>"no physical memory" and type ^t^t k.

Yeah!
^t^t k is a familiar command these days after I began to touch notebook
and kfs.  :-)

On the contrary, we've never experienced in our dual 133MHz petium 
desktop with 48MB memory without local disk (with swapping to other 
filesystem in our fileserver) and connected to networked file server has 
never let us to issue that one.  In this case, however, many heavy works
are doen under the CPU server.  :-)

Kenji

Re: [9fans] python and lua

[Russ Cox]

> What kind of situations?
> I asked this because I've never seen it.

Usually a terminal booted from local disk
can't sustain more than a few minutes of local disk swapping
before the window system or kfs swaps in a bogus page
and dies.  

Try running a few instances of the following program

	for(i in `{seq 10})
		memeat -x 10000000

and sit back.  This doesn't seem to happen when swapping
to a network file server.  Perhaps the timing is different.

I suspect something in the x86 memory management code,
although presotto says the carreras had similar problems.
I spent a day this summer trying to find anything, but
like everyone else came up empty handed.  

Instead, I moved more memory in from another machine
and turned off swapping.  On a laptop, of course,
this is less of an option, but I don't run my laptop
with swapping enabled either.  I'd much rather see 
"no physical memory" and type ^t^t k.

Russ


g% cat memeat.c
#include <u.h>
#include <libc.h>

void
main(int argc, char **argv)
{
	int xflag=0;
	int i, n;
	char *p;

	ARGBEGIN{
	case 'x':
		xflag=1;
	}ARGEND;
	n = atoi(argv[0]);
	p = malloc(n);
	memset(p, 0, n);
	i=0;
	for(;;) {
		sleep(1000+1000*(rand()%8));
		if(xflag)	memset(p, i++, n);
	}
}
g%

Re: [9fans] python and lua

[forsyth]

>>I should have been more specific ont his problem.   I guess this is concerned
>>with memory swapping on a swap partition of local disk same with that kfs system 
>>residing.

we see problems even with a real file server, and suspect
a bug in the paging code but haven't yet found it
(which might simply mean the cause is somewhere else).

Re: [9fans] python and lua

[forsyth]

>>I'm running a local kfs, would these go away with a real
>>fileserver?

in a very useful sense, nothing goes away with a real file server
(with a worm)

Re: [9fans] python and lua

[okamoto]

No!
I forgot to write one line

>--rw-r----- S 0 okamoto okamoto       12354048 Jan 5 09:47 /dev/sdC0/dos
>--rw-r----- S 0 okamoto okamoto  1428553728 Jan 5 09:47 /dev/sdC0/plan9
>-lrw------- S 0 okamoto okamoto                       0 Jan 5 09:47 /dev/sdC0/raw

Those lines should be

--rw-r----- S 0 okamoto okamoto       12354048 Jan 5 09:47 /dev/sdC0/dos
--rw-r----- S 0 okamoto okamoto  1388203008 Jan 5 09:47 /dev/sdC0/fs
--rw-r----- S 0 okamoto okamoto  1428553728 Jan 5 09:47 /dev/sdC0/plan9
-lrw------- S 0 okamoto okamoto                       0 Jan 5 09:47 /dev/sdC0/raw

Re: [9fans] python and lua

[okamoto]

[-- Attachment #1: Type: text/plain, Size: 1317 bytes --]

>Kfs is not so robust as real fileserver.  I'm now facing this many times.

I should have been more specific ont his problem.   I guess this is concerned
with memory swapping on a swap partition of local disk same with that kfs system 
residing.

My /dev/sdC0 directory shows as:

--rw-r----- S 0 okamoto okamoto      10486784 Jan 5 09:47 /dev/sdC0/9fat
--rw-r----- S 0 okamoto okamoto                      0 Jan 5 09:47 /dev/sdC0/ctl
--rw-r----- S 0 okamoto okamoto  1440940032 Jan 5 09:47 /dev/sdC0/data
--rw-r----- S 0 okamoto okamoto       12354048 Jan 5 09:47 /dev/sdC0/dos
--rw-r----- S 0 okamoto okamoto  1428553728 Jan 5 09:47 /dev/sdC0/plan9
-lrw------- S 0 okamoto okamoto                       0 Jan 5 09:47 /dev/sdC0/raw
--rw-r----- S 0 okamoto okamoto       29863936 Jan 5 09:47 /dev/sdC0/swap

I'm using this 30MB swap partition for swap region when I use kfs filesystem.

When I tried to compile 9pcdisk with rio, ktrans etc (full spec of my machine),
link loader may face dead lock, and then, I have to type ^^t on the way of 
linking.
Another more occuring situation is tar-ing a large compressed file such as linux 
kernel source to my home directory.

When I load root filesystem from our fileserver for this same machine, on the contrary, 
I have no probelem.

Kenji


[-- Attachment #2: Type: message/rfc822, Size: 2004 bytes --]

From: okamoto@granite.cias.osakafu-u.ac.jp
To: 9fans@cse.psu.edu
Subject: Re: [9fans] python and lua
Date: Fri, 5 Jan 2001 12:44:31 0900
Date: Fri, 5 Jan 2001 12:44:31 0900
Message-ID: <20010105034447.E920B199D7@mail.cse.psu.edu>

>I'm running a local kfs, would these go away with a real
>fileserver?

Kfs is not so robust as real fileserver.  I'm now facing this many times.
This is my first time to use kfs efficiently, and got a different view on
Plan 9.  :-)  (I'm kidding)

However, I found sam runnes without rio.  Why I know that?  Because
I have to set an environment for sys user without rio etc. to compile
pcdisk kernel in Toshiba Tecra 48MB memory, local swap system.

Kenji

Re: [9fans] python and lua

[okamoto]

Yes, I thought I had known it.
When I saw it in my eye, however, I was surprised.

Kenji

Re: [9fans] python and lua

[rob pike]

	However, I found sam runnes without rio.

Of course it does.  Rio is just a multiplexer of its own environment.

-rob

Re: [9fans] python and lua

[okamoto]

>I'm running a local kfs, would these go away with a real
>fileserver?

Kfs is not so robust as real fileserver.  I'm now facing this many times.
This is my first time to use kfs efficiently, and got a different view on
Plan 9.  :-)  (I'm kidding)

However, I found sam runnes without rio.  Why I know that?  Because
I have to set an environment for sys user without rio etc. to compile
pcdisk kernel in Toshiba Tecra 48MB memory, local swap system.

Kenji

[9fans] python and lua

[Quinn Dunkan]

> I'd like to know who else is interested in this and if anyone is
> interested if we could get a Plan 9 Python project together.  I would
> like to see Plan 9 support, native (though with sockets to support
> various existing internet client classes) and with classes to support
> Plan 9's unique features, in some future release of Python.
> 
> I've made a start at getting it going, and I can put in some more
> work, but I can't do most of it; once the next term starts, I'll have
> to concentrate on my (non-CS) schoolwork.  So are there enough of us
> with interest and time to keep at it?

I'm interested in this too.  I also made a port, but ran into trouble when
trying to write extension modules.  The problem was that I needed to compile
python under the APE, but extension modules had to be compiled natively to use
the native plan9 functions.  I got tired of monkeying around with APE
impedance-matching, and decided to port lua to compile natively, since lua has
fewer posix dependencies and is smaller in general.  Besides, I realized that
all my plan9 hack time was going into learning the vagaries of the plan9
emulation of the same old posix, which is sort of missing the point :)

So I wrote out a mkfile and hacked around all the lua posixlyness, and now I'm
starting on a p9lib wrapper which will provide the basic syscalls.  I might
write an io library that uses BIO instead of stdio, and purge stdio from the
interpreter, which ought to make it a bit smaller, but mostly give me an
excuse to get familiar with BIO.  Then I'll see about writing a lua library to
speak 9p, so I can easily set up "language driven fileservers" like the
plumber, even on non-plan9 systems (provided 9p will work over tcp...).  Maybe
experiment with bouncing lua bytecode around with the plumber, if I can think
of any sort of use for that.

Anyway, seeing as python is still one of my favorite languages, I'd love to
see it under plan9 as well.  And plan9 definitely needs a few more high-level
languages (I mean, awk is great and all, but... you know...).



And some miscellanea---miscellaneou---ARGH!  Some random questions:

Are the plan9 fonts anti-aliased?  It sorta looks like they are, but that may
be the result of bad eyes, a dusty monitor, and a fuzzy voodoo3 card.

I get lots of random slow-downs, where entering text or something will hang
for just a moment.  I can get quite long hangs if someone sucks up cpu time,
like if I ask 'page' to reverse its image.  Especially scrolling text in acme
is slow---if I click on the scrollbar at a steady rate, it will scroll in
irregular bursts.  Is the voodoo3 driver accelerated?  I do have a 'echo accel
on >#v/vgactl' in my termrc, but it doesn't seem to make a lot of difference.
Is this usual?  I'm running a local kfs, would these go away with a real
fileserver?

Re: [9fans] python and lua

[rob pike]

[-- Attachment #1: Type: text/plain, Size: 113 bytes --]

It's a no-op on the 3dfx, I believe.  No one has written the
interace code for the hardware support.

-rob


[-- Attachment #2: Type: message/rfc822, Size: 1410 bytes --]

From: arisawa@ar.aichi-u.ac.jp
To: 9fans@cse.psu.edu
Subject: Re: [9fans] python and lua
Date: Fri,  5 Jan 101 11:50:11 +0900
Message-ID: <20010105032940.693B0199D7@mail.cse.psu.edu>

Hello,

>I do have a 'echo accel
>on >#v/vgactl' in my termrc, but it doesn't seem to make a lot of  
>difference.
Try:
echo -n 'hwaccel on' >'#v/vgactl'

Kenji Arisawa
E-mail: arisawa@aichi-u.ac.jp

Re: [9fans] python and lua

[arisawa]

Hello,

>I do have a 'echo accel
>on >#v/vgactl' in my termrc, but it doesn't seem to make a lot of  
>difference.
Try:
echo -n 'hwaccel on' >'#v/vgactl'

Kenji Arisawa
E-mail: arisawa@aichi-u.ac.jp

Re: [9fans] python and lua

[rob pike]

There are a few antialiased fonts, all derived from outline.

	echo /lib/font/bit/*/*.1

for ldepth-1 subfont files.

Asking page to reverse its image is one place in which the graphics
model gets in the way; the system gets very busy nattering back and
forth between the application and the graphics driver.  A similar
thing happens when loading fonts; on a slow line, rio takes a while
to respond when you cat a Japanese file.  The reason for your
`lots of random slow-downs' eludes me, though.  I do think that
getting hardware acceleration running on the voodoo3 (I'm pretty
sure no one's done the work) will help somewhat.  The system 
will surely feel zippier.  Anyone?  It's easy if you have the spec.

-rob