Re: Something stopping Gimp loading

[Joost Witteveen]

> [EMAIL PROTECTED]:~$ /sbin/ldconfig -pNX | grep libglib-2.0
>libglib-2.0.so.0 (libc6,x86-64) => /usr/local/lib/libglib-2.0.so.0
>libglib-2.0.so.0 (libc6,x86-64) => /usr/lib/libglib-2.0.so.0
>libglib-2.0.so (libc6,x86-64) => /usr/local/lib/libglib-2.0.so
> [EMAIL PROTECTED]:~$ ldd `which gimp` \  [K| grep libglib-2.0
>libglib-2.0.so.0 => /usr/local/lib/libglib-2.0.so.0 
> (0x2b5b4ecea000)

So, somehow you've installed another libglib-2.0 in /usr/local, and I
suppose it's using that. Remove the one in /usr/local/lib, and I
suppose everything will be fine again.


-- 
To UNSUBSCRIBE, email to [EMAIL PROTECTED] 
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]

Re: undefined reference to `gettid'

[Joost Witteveen]

2008/4/28 Mathieu Malaterre <[EMAIL PROTECTED]>:
> On Mon, Apr 28, 2008 at 5:24 PM, Eduardo M KALINOWSKI <[EMAIL PROTECTED]> 
> wrote:
>  > Mathieu Malaterre wrote:
>  >  > Should I simply assumed that gettid should not be used and instead
>  >  > replaced it with pthread_self as this should be the default on all
>  >  > linux/debian system ?
>  >  >
>  >
>  >  >From man gettid(2):
>  >
>  >  NOTES
>  >Glibc does not provide a wrapper for this system call;  call  it
>  >  using
>  >syscall(2).
>  >
>  >The  thread  ID  returned by this call is not the same thing as a
>  >  POSIX
>  >thread ID (i.e., the opaque value returned by pthread_self(3)).
>  >
>  >
>  >  However, concerning if you *should* call it via syscall, I cannot help.
>  >
>
>  WTF ? I am running an up to date debian stable system, where are those
>  new man pages from ?

[EMAIL PROTECTED]:/$ dpkg -S /usr/share/man/man2/gettid.2.gz
manpages-dev: /usr/share/man/man2/gettid.2.gz

[EMAIL PROTECTED]:/$ dpkg -l manpages-dev
Desired=Unknown/Install/Remove/Purge/Hold
| Status=Not/Installed/Config-files/Unpacked/Failed-config/Half-installed
|/ Err?=(none)/Hold/Reinst-required/X=both-problems (Status,Err: uppercase=bad)
||/ NameVersion Description
+++-===-===-==
ii  manpages-dev2.39-1  Manual pages about using
GNU/Linux for development

Where does your gettid manpage come from?


-- 
To UNSUBSCRIBE, email to [EMAIL PROTECTED] 
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]

Re: pdf plugin

[Joost Witteveen]

On 21/04/2008, Alex Samad <[EMAIL PROTECTED]> wrote:
> Hi
>
>  what package do I need to install to get the firefox pdf plugin so I can
>  view pdf inline ?

mozplugger, from the main debian archive:

Description: Plugin allowing external viewers to be launched inside Mozilla
 mozplugger allows you to seamlessly integrate external applications
 to view files downloaded from the web that Mozilla can not normally
 handle. The application is embedded within a Mozilla window as to act
 like and feel like a true plugin.
 .
 This allows to you view PDFs, Postscript files, animations and
 movies, amongst other file types all from within Mozilla (with
 supporting applications).


I just installed it, and now firefox shows pdf's inline. I'll
uninstallit again, as I don't like that, but it does work:).


-- 
To UNSUBSCRIBE, email to [EMAIL PROTECTED] 
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]

Re: Source code editor

[Joost Witteveen]

On 21/04/2008, Johann Spies <[EMAIL PROTECTED]> wrote:
> On Fri, Apr 18, 2008 at 12:12:40PM +0200, Lars Bjerregaard wrote:
>  > Jordi Gutiérrez Hermoso wrote:
>  >> On 16/04/2008, Tero Mäntyvaara <[EMAIL PROTECTED]> wrote:
>  >>>  I installed motor in my AMD64 Etch system and after I execute it I get
>  >>> "Segmentation fault"... :-/
>  >>
>  >> Whoa, etch? You got that on etch? How the hell did such an RC bug get
>  >> into etch? First time I hear about it.
>
>
> I also get a 'Segmentation fault' on a 64-bit system.

Me too, it needs (at least) the following patch:

diff -u -r motor-3.4.0/src/ui/ncurses/ncursesui.cc
motor-3.4.0-patch/src/ui/ncurses/ncursesui.cc
--- motor-3.4.0/src/ui/ncurses/ncursesui.cc 2008-04-21
10:42:24.0 +0200
+++ motor-3.4.0-patch/src/ui/ncurses/ncursesui.cc   2008-04-21
10:32:51.0 +0200
@@ -338,7 +338,7 @@

 if(amode == selectorcreate) {
db.setbar(new horizontalbar(getcolor(cp_menu), getcolor(cp_menusel),
-   _("Create/Import.."), _("Add"), _("Remove"), _("Open"), 0));
+   _("Create/Import.."), _("Add"), _("Remove"), _("Open"), (char *)0));

db.addkey(KEY_IC, 1);
db.addkey(KEY_DC, 2);

On AMD64 bit, int is 32 bit, while the pointers are 64 bits. So when
it pushes the int(0) as last argument on the stack for
horizontalbar(nc,sc, ...), it only pushes 4 bytes worth of 0's, in
stead of 8.
Problem is, I see a lot of functions using "..." as argument, and
cannot easily find where they are used to make sure there are no other
instances of this.

At least with the patch above it starts on my AMD 64 bit system.
I'll file a bug for it on motor.

Re: removed root directory files

[Joost Witteveen]

On 17/04/2008, Abraham Chaffin <[EMAIL PROTECTED]> wrote:
> Hey All,
>
> Bit of a problem here - I ran rm /*
> It didn't remove the directories obviously but now no commands work but cd

cd is a bash-builtin, that's why it works. "echo", "type", "cat", etc
also work, I suppose.

> I was wanting to run ./*
> Now if I ls I get
> -su: /bin/ls: No such file or directory

I suppose the /bin/ls file still is OK (test by "cd /bin; echo *", you
should see ls as one of the many commands listed), but either
/lib/ld-linux.so or some other library has gone. That is strange, as
just typing "rm /*" should not remove anything but files in / , so I
suspect you actually typed something else than "rm /*"

>  Help much appreciated.

Insert debian boot CD, boot, mount your root partition of harddisk,
and inspect it.
Maybe you only removed a few files from /lib, you might be able to
copy them from the boot disk /lib. But I suppose a re-install of
Debian would be safer (or wait till someone else has a better idea on
this list).


-- 
To UNSUBSCRIBE, email to [EMAIL PROTECTED] 
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]

Re: Problem compiling simple C program

[Joost Witteveen]

On 16/04/2008, John Salmon <[EMAIL PROTECTED]> wrote:
> Sven Joachim <[EMAIL PROTECTED]> wrote in news:[EMAIL PROTECTED]:
>
>
>  > On 2008-04-15 20:39 +0200, John Salmon wrote:
>  >
>  >> I'm running Debian Etch on a PC. When I try to compile the following
>  >> (called test.c);
>  >>
>  >> #include 
>  >> #include 
>  >>
>  >> int main()
>  >> {
>  >> double
>  >> val = 1.55;
>  >>
>  >> printf("sine: %g\n", sin(val));
>  >>
>  >> return 0;
>  >> }
>  >>
>  >> using the command line
>  >>
>  >> gcc -Wall -o test test.c
>  >>
>  >> I get
>  >>
>  >> /tmp/cciDV02m.o: In function `main':
>  >> test.c:(.text+0x21): undefined reference to `sin'
>  >> collect2: ld returned 1 exit status
>  >
>  > You need to link to the math library by specifying -lm _at the end_ of
>  > the gcc command line, otherwise the linker does not know about the sin()
>  > function.
>  >
>  >> When I compile the equivalent C++ program using the apropriate C++
>  >> parameters, everything goes great. Have I neglected to load a Debian
>  >> package? Any help will be appreciated.
>  >
>  > That is to be expected, because C++ programs are automatically linked
>  > against the math library.  In C you have to tell the linker to use it
>  > with -lm.
>  >
>  > Sven
>  >
>  >
>
>
> That solved the linking problem. Now, after a successful compilation, when
>  I run 'test' I gen no output.   ???


"test" is a bash-builtin, and is a executable in /usr/bin/test, so
that is what you were running.
You can run your "test" command by typing
./test
or rename it to something else.


-- 
To UNSUBSCRIBE, email to [EMAIL PROTECTED] 
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]

Re: grep trick

[Joost Witteveen]

On 09/04/2008, Chris Bannister <[EMAIL PROTECTED]> wrote:
> On Tue, Apr 08, 2008 at 10:22:08AM +1000, Rich Healey wrote:
>  > -BEGIN PGP SIGNED MESSAGE-
>  > Hash: SHA1
>  >
>  > Mike Bird wrote:
>  > > On Mon April 7 2008 16:03:28 Chris Bannister wrote:
>  > >> export GREP_COLOR=33
>  > >> alias grep='grep --colour=always'
>  > >
>  > > This will break any scripts which assume that the output
>  > > of grep has not been annotated with color escape sequences.
>  > >
>  > > --Mike Bird
>  > >
>  > >
>  > Those scripts will not load his .bashrc
>
>
> Which ones? e.g ... as me:
>
>  echo $GREP_COLOR
>  33
>
>  less gtt
>  #!/bin/bash
>
>  echo $GREP_COLOR
>
>  ./gtt
>  33
>
>  So any scripts run *as* me get it too.

More precicely, any scripts you run from interactive sessions (terminal etc).
Scripts started under your UID from cron, at, etc, will not get the
GREP_COLOR var.


-- 
To UNSUBSCRIBE, email to [EMAIL PROTECTED] 
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]

Re: error while loading shared libraries: libstdc++.so.5

[Joost Witteveen]

On 07/04/2008, Haines Brown <[EMAIL PROTECTED]> wrote:
> "Joost Witteveen" <[EMAIL PROTECTED]> writes:
>
>  >>  > ldd /usr/bin/gmfsk
>  >>
>  >>
>  >>   $ ldd /usr/bin/gmfsk | grep libstdc
>  >> libstdc++.so.6 => /usr/lib/libstdc++.so.6 (0xb738e000)
>  >>
>  >>  So, since I had libstsdc++.so.6 installed, and this is what it needs,
>  >>  why would installation of libstdc++.so.5 enabled it to work?
>  >
>  > If it were to happen to me, I'd assume it was because I didn't look
>  > properly when I tried it without libcstc++.so.5 (and I'd try removing
>  > the libstdc++5 version to verify).
>  > But I won't assume you're as stupid as me, so maybe gmfsk executes
>  > another binary that depends on libstdc++.so.5. To try that, uninstall
>  > libstdc++.so.5, and run
>  > strace -f gmfsk 2> stderr-file.txt
>  > then search the stderr-file.txt for libstdc. If there is something
>  > opening libstdc++.so.5, you should see the exec() line somewhere
>  > above.
>
>
> Interesting, I couldn't remove libstdc++.so.5:
>
>  $ ls -l /usr/lib | grep libstdc++.so*
>  lrwxrwxrwx  1 root root   18 2008-04-07 06:14 libstdc++.so.5 ->
>
> libstdc++.so.5.0.7
>
> -rw-r--r--  1 root root   737624 2007-01-03 13:47 libstdc++.so.5.0.7
>  lrwxrwxrwx  1 root root   18 2007-04-22 09:40 libstdc++.so.6 ->
>
> libstdc++.so.6.0.8
>
> -rw-r--r--  1 root root   909044 2006-12-10 09:45 libstdc++.so.6.0.8
>
>  $ sudo aptitude remove libstdc++.so.5.0.7

$ dpkg -S libstdc++.so.5.0.7
libstdc++5: /usr/lib/libstdc++.so.5.0.7

So the package is called libstdc++5.

On my system not much depends on it, only lsb and g++-3.3 (old).


-- 
To UNSUBSCRIBE, email to [EMAIL PROTECTED] 
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]

Re: error while loading shared libraries: libstdc++.so.5

[Joost Witteveen]

>  > ldd /usr/bin/gmfsk
>
>
>   $ ldd /usr/bin/gmfsk | grep libstdc
> libstdc++.so.6 => /usr/lib/libstdc++.so.6 (0xb738e000)
>
>  So, since I had libstsdc++.so.6 installed, and this is what it needs,
>  why would installation of libstdc++.so.5 enabled it to work?

If it were to happen to me, I'd assume it was because I didn't look
properly when I tried it without libcstc++.so.5 (and I'd try removing
the libstdc++5 version to verify).
But I won't assume you're as stupid as me, so maybe gmfsk executes
another binary that depends on libstdc++.so.5. To try that, uninstall
libstdc++.so.5, and run
strace -f gmfsk 2> stderr-file.txt
then search the stderr-file.txt for libstdc. If there is something
opening libstdc++.so.5, you should see the exec() line somewhere
above.


-- 
To UNSUBSCRIBE, email to [EMAIL PROTECTED] 
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]

Re: localtime and UTC

[Joost Witteveen]

On 04/04/2008, Luca Sighinolfi <[EMAIL PROTECTED]> wrote:
> On Fri, 4 Apr 2008 17:51:34 +0800
>
> "Rage Callao" <[EMAIL PROTECTED]> wrote:
>
>
> > On Fri, Apr 4, 2008 at 3:20 PM, Luca Sighinolfi
>  > <[EMAIL PROTECTED]> wrote:
>
> > >  Well, I think this is normal if you have set a time zone!
>  >
>  > Hi. Do you mean that the Lenny installer expects me not to set a time
>  > zone if my machine's BIOS clock is set to localtime?
>
>
> The answer is yes: Debian expects your BIOS is set on UTC.

But if have a dual boot system with the other system expecting BIOS
set to localtime, Debian should be able to handle this:

http://www.debian.org/doc/manuals/system-administrator/ch-sysadmin-time.html#s-multiboot-with

It may cause problems when rebooting after sommer/winter time changes
though, as Debian has no way of knowing whether the other system
already ajusted the local clock to the new sommer/winter time.

(To set hwclock to UTC/localtime, edit /etc/default/rcS)


-- 
To UNSUBSCRIBE, email to [EMAIL PROTECTED] 
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]

Re: Display managers fail since latest upgrade to Lenny

[Joost Witteveen]

On 03/04/2008, Sebastian Tennant <[EMAIL PROTECTED]> wrote:
> Hi list,
>
>  Since I last upgraded my Lenny box, both xdm and gdm silently fail to
>  start at boot time.
>
>  Actually, something in each of their init scripts fails and they are
>  never actually called.

you could type

sh -x /etc/init.d/gdm start

to see what gets executed, and maybe have a better idea of what causes
the problem.


-- 
To UNSUBSCRIBE, email to [EMAIL PROTECTED] 
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]

Re: Hmmm. A question. Was [Re: Debian is losing its users]

[Joost Witteveen]

On 30/03/2008, Owen Townend <[EMAIL PROTECTED]> wrote:
[...]
> They also alphabetically increment with each release and Gutsy Gibbon,
> Hardy Herron
Going back to google trends, these are nice graphs:

http://www.google.com/trends?q=dapper+drake+%2C+Edgy+Eft%2C+Feisty+Fawn%2C+Gutsy+Gibbon%2C+Hardy+heron

And going back to Debian:
http://www.google.com/trends?q=debian+woody%2C+debian+sarge%2C+debian+etch%2C+debian+lenny


-- 
To UNSUBSCRIBE, email to [EMAIL PROTECTED] 
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]

Re: openssh 4.7p1 Debian-4, OpenSSL 0.9.8g hangs after authentication

[Joost Witteveen]

On 27/03/2008, Eike Lantzsch <[EMAIL PROTECTED]> wrote:
> Hi:
>
>  Does somebody have a hint for me how to investigate this problem further:
>  Openssh hangs from my workstation to any ssh-server including the workstation
>  itself.
>  Using another client machine I can ssh into all machines allright, including
>  the Fawlty [(c)John Cleese] one. One has also installed Debian testing with
>  the very same version of openssh and openssl.
>
>  [EMAIL PROTECTED]:~$ ssh -vv -4 -2 -l myuser 192.168.0.40
[...]
>  debug2: service_accept: ssh-userauth
>  debug1: SSH2_MSG_SERVICE_ACCEPT received

I'd run strace on it.
When I do
  strace -f ssh -vv -4 -2 -l joostje muso.komputilo.org 2>strace.txt
then I see in strace.txt:
[...]
write(2, "debug2: service_accept: ssh-user"..., 38debug2:
service_accept: ssh-userauth
) = 38
write(2, "debug1: SSH2_MSG_SERVICE_ACCEPT "..., 42debug1:
SSH2_MSG_SERVICE_ACCEPT received
) = 42
write(2, "debug2: key: /home/joostje/.ssh/"..., 48debug2: key:
/home/joostje/.ssh/id_rsa ((nil))
) = 48
write(2, "debug2: key: /home/joostje/.ssh/"..., 51debug2: key:
/home/joostje/.ssh/id_dsa (0x55dff0)
) = 51
write(3, "o\243\302\260\251\5\264\215o\'}3N\341\354\225\'\215\213"..., 64) = 64
select(4, [3], NULL, NULL, NULL)= 1 (in [3])
read(3, "d\265\232\322\'\v\215Md\370.t\2511\353\274\267G\212L\302"...,
8192) = 64
write(2, "debug1: Authentications that can"..., 63debug1:
Authentications that can continue: publickey,password
) = 63
[...]
In other words, my ssh writes two more debug statements right after
the last debug statements you see, without any system calls in
between.

So, basically, it seems like it is 'impossible' for ssh to hang just
at that point...
Would be interesting to see the strace output of your ssh.

After that it starts to write to fd 3, witch seems to be the
communication socket to the SSH server.

My versions:
ii  openssh-client   4.3p2-9  Secure shell
client, an rlogin/rsh/rcp replacement
ii  openssh-server   4.3p2-9  Secure shell
server, an rshd replacement
ii  openssl  0.9.8c-4etch1Secure Socket
Layer (SSL) binary and related cryptographic tools


-- 
To UNSUBSCRIBE, email to [EMAIL PROTECTED] 
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]

Re: Tunnel iceweasel?

[Joost Witteveen]

> Hi, the issue here isn't the speed

Well, you mentioned "it takes forever", so I thought I'd help you with
speed as well:).

>  and besides, i prefer to have it
>  directly connected to my Xserver, rather than runnign in VNC.

Me too, but for me a direct X connection is simply to slow.

>  The point here isn't eh startup time though, it's that it starts a local
>  iceweasel!
>
>  In trying to build FF from source on my new 64 bit machine i
>  accidentally wound up with a ff3 beta, but running that now also opens
>  iceweasel.
>
>  Somehow the binary has managed to associate EVERYTHING with itself.
>
>  The real thing that does my head in is when i launch FF on another box..
>  it still creates a local iceweasel? this should happen AFAIK..

I know, and I'm often annoyed by it as wel.

>  starting a command on that box via should not be able to cause commands
>  to be run on my local?

It is trying to be helpfull, of cource. When I click on a link in an
email, I'd want the page to appear in my already running firefox
session. So it always checks to see if a firefox session is running on
the X display, and if so, it will instruct that firefox session to
display the page.

>  Does this constitute a security issue? i'll see if i can get a PoC
>  during the week, even if one couldn't get arbitrary code, one could
>  still point the new iceweasel on the host machine to a site witha FF
>  exploit.

If you allow a sever to connect to your X display, you allow it to
view all keypresses, what programs are running, what your display
looks like, etc. As far as I know, the only way to avoid that is,
well, to use for example VNC:).

It's not a Firefox problem anyway, it's the way X was designed.


-- 
To UNSUBSCRIBE, email to [EMAIL PROTECTED] 
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]

Re: migrating to 64 bit...

[Joost Witteveen]

On 26/03/2008, David Fox <[EMAIL PROTECTED]> wrote:
> On Tue, Mar 25, 2008 at 3:44 PM, Andrew Sackville-West >  That means I
>
> get to move up to 64 bit. In keeping with my personal
>
>  >  preference to *never* reinstall, I've got an opportunity to attempt to
>  >  migrate a running system from 32 to 64 bit. I also have the
>  >  opportunity to practice on my laptop which could run 64 bit but
>
>
> Here's a thought. If you can use the laptop, go ahead, but can you
>  carve out a small place on your desktop for this? I suggest basically
>  doing a new partition, doing a debootstrap of the current version of
>  debian you already use, for the amd64 architecture. Prior to doing the
>  debootstrap, use dpkg --set-selections and save that in a convenient
>  place, then do a dpkg --get-selections to get all the packages you
>  already have.

I'd use the two commands the other way around:

  dpkg --get-selections [package-name-pattern...]
 Get list of package selections, and write it to stdout.  Without
 a pattern, packages marked with state purge will not be shown.

  dpkg --set-selections
 Set  package  selections  using  file read from stdin. This file
 should be in the format ' ', where state is  one
 of  install,  hold,  deinstall or purge. Blank lines and comment
 lines beginning with '#' are also permitted.


But yes, that seems the most sensible way to do it, though it seems OP
isn't really looking for the most sensible way:). I would assume that
with some dpkg abuse it should be possible to install a 64 kernel
(that also can run 32 userland), then maybe copy /lib and /usr/lib to
/lib32 and /usr/lib32 (and add those locations to /etc/ld.so.conf.d),
abuse dpkg some more that it starts installing 64 bit libraries and
executables, and ...
At some points in between you may well end up with a hosed system, but
well, if you don't mind that, it seems like an interesting route:).


-- 
To UNSUBSCRIBE, email to [EMAIL PROTECTED] 
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]

Re: Errant ld-linux.so.2

[Joost Witteveen]

On 25/03/2008, Patrick Wiseman <[EMAIL PROTECTED]> wrote:
> Every now and then, I have an apparently memory-leaking ld-linux.so.2
>  process running.  When I kill it, nothing else seems adversely
>  affected.  Is there some way I can identify what's spawned it?

Normally you would be able to see what process spawned what using
  ps axf

(f:   ASCII-art process hierarchy (forest))
But I suppose in your case you may not see much interesting output.
In that case, I'd write a short shell script that does ps ax>file
continually, to try and see when the process starts to appear.


-- 
To UNSUBSCRIBE, email to [EMAIL PROTECTED] 
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]

Re: Tunnel iceweasel?

[Joost Witteveen]

On 24/03/2008, Douglas A. Tutty <[EMAIL PROTECTED]> wrote:
> On Mon, Mar 24, 2008 at 11:46:56AM +0100, Joost Witteveen wrote:
>  > On 23/03/2008, Rich Healey <[EMAIL PROTECTED]> wrote:
>
> > > I'm trying to tunnel an iceweasel instance via ssh from one
>  > > of my boxes at my house to remember the name of an add-on i
>  > > installed.
>  > >
>  > > The problem is that i create a ssh session (ssh -XC
>  > > ssh.psychotik.info), login and run iceweasel at the bash
>  > > prompt, which takes forever, but then finally *opens a local
>  > > iceweasel!!!*
>  >
>  > I suppose that iceweasel -P uniqueprofilename would do what you want?
>  >
>  > Also, it's *much* faster use vnc (tunnel through ssh): on the remote
>  > host, start: vnc4server on your localhost, start (and login to) ssh -L
>  > 5900:server:5901 server
>  >
>  > and then on the localhost (different window) vncviewer localhost:5900
>  >
>  > The 5901 portnumer is assuming the vncserver opens a X11 screen on :1.
>  > When I start epiphany diretly over X11, it takes about 30 min to show
>  > a page; when I do it using VNC as above, it takes seconds.
>
>
> I run iceweasel over ssh all the time, however, I don't have it
>  installed locally so there's no local version to run.  It may take a few
>  seconds to give the initial window, but then it displays as fast as the
>  box can swap.  The network is 100 MB/s ethernet, the box I'm sitting at
>  is a P-II with 64 MB ram, the box I'm sshing into to run iceweasel is an
>  AMD Athlon64 with 1 GB ram.  It doesn't even take 30 minutes to show a
>  page when I ssh from my 486 with 32 MB ram so something is wrong there.
>
>  Why would VNC be faster if both are encrypted?
>

No, over a 100Mb/s ethernet, running iceweasel over VNC probably
wouldn't be much faster than directly over ssh (and running over an
ssh-tunneled VNC connection would of course be slower than straigt
VNC).

But the OP complained iceweasel was very slow. So I suppose he didn't
run it over a direct 100Mb/s connection, but over something slower,
probably with larger ping times, ping times of 10-30 ms are enough to
make it slow, and with slow, I mean that it can take over 20 min for
iceweasel to even start showing the home page.
I notice that when that happens, starting iceweasel on the remote site
on a VNC X server an watching the output via a VNC viewer is a lot
faster. And a lot here means just a couple of seconds to show the home
page, instead of 20 min.
As the OP reported using ssh, I assumed he didn't want to connect
unencrypted (somethign VNC as far as I know does), so I suggested
using an ssh tunnel.


-- 
To UNSUBSCRIBE, email to [EMAIL PROTECTED] 
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]

Re: Upgrading Sarge to Etch Pining apache 1.3 and php4

[Joost Witteveen]

On 24/03/2008, Siju George <[EMAIL PROTECTED]> wrote:

>  I have a Few Server running Apache 1.3 and PHP 4
>  Is it enough that I put in /etc/apt/preferences
>
>  Package: apache
>  Pin: version 1.3.*
>
>  Package: php4
>  Pin: version 4.*
>
>  and edit sarge from mt /etc/apt/source.list to put etch instead there after 
> do
>
>  apt-get update
>  and
>  apt-get dist-upgrade

I'm not sure what your question is (didn't see that email) but if you
just want to upgrade to etch and stay with apache 1.3 & php 4, then
you can just go ahead and upgrade to etch without any pinning of
apache/php versions.
The apache 2.2 and php 5 packages in etch are called "apache2" and
"php5" and they will not automatically replace your "apache" (1.3) and
"php4" packages.

>  Anything else I need to keep in mind during the upgrade?

probably depends on what you are doing with it, but I don't think
anything will go wrong that you cannot fix with ssh.


-- 
To UNSUBSCRIBE, email to [EMAIL PROTECTED] 
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]

Re: Configuring dhcp3 to send hostname mappings to clients

[Joost Witteveen]

On 23/03/2008, Reid Priedhorsky <[EMAIL PROTECTED]> wrote:
> Dear all,
>
>  My /etc/hosts contains a handful of mappings useful on my internal subnet:
>
>  10.0.0.2  reidster.net reidster
>  10.0.0.4  convex.reidster.net  convex
>  10.0.0.4  toys.reidster.nettoys.reidster
>  10.0.0.4  yabman.reidster.net  yabman
>
>  The same box is also a DHCP server, using dhcp3 for a couple of laptops
>  which are transient on the network. I would like to pass the mappings
>  above on to the laptops using dhcp3. Is this possible?

I don't think it is.

What I would do is install a DNS server (eg bind9), and in
/etc/dhcpd.conf put a line like:
  option domain-name-servers 10.0.0.4;
and then configure the bind to serve the mappings you want.
(you'll need seperate files for reverse mappings)


-- 
To UNSUBSCRIBE, email to [EMAIL PROTECTED] 
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]

Re: Tunnel iceweasel?

[Joost Witteveen]

On 23/03/2008, Rich Healey <[EMAIL PROTECTED]> wrote:
> -BEGIN PGP SIGNED MESSAGE-
>  Hash: SHA1
>
>  Hi list,
> I'm trying to tunnel an iceweasel instance via ssh from one of my 
> boxes
>  at my house to remember the name of an add-on i installed.
>
> The problem is that i create a ssh session (ssh -XC
>  ssh.psychotik.info), login and run iceweasel at the bash prompt, which
>  takes forever, but then finally *opens a local iceweasel!!!*

I suppose that
 iceweasel -P uniqueprofilename
would do what you want?

Also, it's *much* faster use vnc (tunnel through ssh):
on the remote host, start: vnc4server
on your localhost, start (and login to)
 ssh -L 5900:server:5901 server

and then on the localhost (different window)
 vncviewer localhost:5900

The 5901 portnumer is assuming the vncserver opens a X11 screen on :1.
When I start epiphany diretly over X11, it takes about 30 min to show
a page; when I do it using VNC as above, it takes seconds.


-- 
To UNSUBSCRIBE, email to [EMAIL PROTECTED] 
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]

Re: SWIG, STL, Python: std_common.i:9: Error: Syntax error in input(1).

[Joost Witteveen]

>  [EMAIL PROTECTED]:~/ar$ swig -python example.i
>  /usr/share/swig1.3/python/std_common.i:9: Error: Syntax error in input(1).

At least when giving the -c++ argument to swig it works better:

[EMAIL PROTECTED]:~/ar$ swig -python -c++ example.i

doesn't look ideal (should accept same arguments for all languages, I
guess), but at least it works.


-- 
To UNSUBSCRIBE, email to [EMAIL PROTECTED] 
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]

SWIG, STL, Python: std_common.i:9: Error: Syntax error in input(1).

[Joost Witteveen]

I try the SWIG STL vector example from:

http://www.swig.org/Doc1.3/Library.html#Library_nn15

calling swig seems to work for all languages, except for python (the
one I want):

[EMAIL PROTECTED]:~/ar$ swig -python example.i
/usr/share/swig1.3/python/std_common.i:9: Error: Syntax error in input(1).

Does anyone know what I can do to make it work?
A google search only showed
http://blog.isnotworking.com/2006/08/word-of-caution-distutils-swig-stl.html
about setting an undocumented swig_opts=['-c++'] option in setup.py, I
don't know what setup.py file that is (and strace shows swig doesn't
look for setup.py files).


Other languages work OK with the same example.[ih] files:

[EMAIL PROTECTED]:~/ar$ swig -tcl example.i
[EMAIL PROTECTED]:~/ar$ swig -perl5 example.i
[EMAIL PROTECTED]:~/ar$ swig -java example.i
[EMAIL PROTECTED]:~/ar$ swig -csharp example.i

Here are the example.i and example.h files, copied from
http://www.swig.org/Doc1.3/Library.html#Library_nn15

[EMAIL PROTECTED]:~/ar$ cat example.i
%module example
%{
#include "example.h"
%}

%include "std_vector.i"
// Instantiate templates used by example
namespace std {
   %template(IntVector) vector;
   %template(DoubleVector) vector;
}

// Include the header file with above prototypes
%include "example.h"
[EMAIL PROTECTED]:~/ar$ cat example.h
/* File : example.h */

#include 
#include 
#include 
#include 

double average(std::vector v) {
return std::accumulate(v.begin(),v.end(),0.0)/v.size();
}

std::vector half(const std::vector& v) {
std::vector w(v);
for (unsigned int i=0; i& v) {
std::transform(v.begin(),v.end(),v.begin(),
       std::bind2nd(std::divides(),2.0));
}

-- 
Thanks,
Joost Witteveen


-- 
To UNSUBSCRIBE, email to [EMAIL PROTECTED] 
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]

Re: daytime

[Joost Witteveen]

[EMAIL PROTECTED] wrote:
/etc/inetd.conf in Sparc Woody contains these lines.
 
 daytime stream  tcp nowait  rootinternal
#daytimedgram   udp waitrootinternal

Anyone have an idea about making this system respond to 
a daytime request from another system on the LAN?
Well, it works for me. Are you sure inetd is running?
What does
  ps `cat /var/run/inetd.pid`
show?
Maybe you are running a firewall that is blocking the requests?
Maybe something in /etc/hosts.alow or /etc/hosts.deny?
--
Groetjes
joostje
47d3fcfe28f2a83497e79d9bc7d5087c-4a1ee1fddab4648175518cb4c1c9edb3ed0e89f0
--
To UNSUBSCRIBE, email to [EMAIL PROTECTED] 
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]

Re: NFS permissions question

[Joost Witteveen]

Christian Convey wrote:
My understanding of NFS permissions is that for any file appearing on an 
NFS share, the username/uid and groupname/gid mappings should (ideally) 
be identical on both the NFS client and the NFS server.

So consider my home situation: I'm running two computers, each with 
local security files.

I have four users: "alvin", "benny", "charles", and "david".
I have several groups: "users" and "chefs" and "busboys".
I want to define an NFS share that "alvin" and "benny" can use. My 
*expectation* at the time I'm setting this up is that any files 
appearing on those shares will have a group-owner of "chefs".

So I go through, and ensure that "alvin" and "benny" each have the same 
uid on both computers. I go through and ensure that "chefs" has the same 
gid on both computers.

Is there a good way for me to ensure that alvin doesn't create, on the 
shares, a file owned by the busboys group?
I probably completely misunderstand what you want, but wouldn't making
the shares owned by group "chefs", and setting the setgid bit on the 
director fix it? (that would probably have to be don on the NFS server)

$ mkdir audio
$ chown joostje:audio audio
$ chmod g+s audio
$ chmod g+w audio
$ cd audio/
$ touch file
$ su otheruser
[...]
$ touch otherfile
$ ls -al
total 8
drwxr-sr-x  2 joostje   audio   4096 2004-11-18 22:58 .
drwxr-xr-x  7 joostje   joostje 4096 2004-11-18 22:58 ..
-rw-r--r--  1 joostje   audio  0 2004-11-18 22:58 file
-rw-r--r--  1 otheruser audio  0 2004-11-18 22:59 otherfile
i.e. both users "joostje" and "otheruser" can write to the dir audio, 
and all files are owned by group "audio".

--
Groetjes
joostje
47d3fcfe28f2a83497e79d9bc7d5087c-4a1ee1fddab4648175518cb4c1c9edb3ed0e89f0
--
To UNSUBSCRIBE, email to [EMAIL PROTECTED] 
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]

Re: Limiting User Commands

[Joost Witteveen]

Michael Graham wrote:
/tmp/test$ ll
total 0
-rw-r--r--  1 root root 0 2004-11-15 00:36 test
/tmp/test$ ll -d ../test/
drwxr-xr-t  2 mick mick 4.0K 2004-11-15 00:36 ../test/
/tmp/test$ rm test
rm: remove write-protected regular empty file `test'? y
/tmp/test$ ll
total 0
But according to the man page of chmod I shouldn't be able to do this:
Interesting. I notice that after a `chown root:root /tmp/test', the file
/tmp/test/test cannot be removed any more by the normal user.
So, I suppose the fact that mick can remove the file /tmp/test/test is 
related to the fact that /tmp/test is owned by mick (or 'joostje' in my 
case).

There is some logic to it, because if the kernel didn't allow mick to 
remove /tmp/test/test, mick could simply first run
  chmod u-t /tmp/test
and then repeat the rm /tmp/test/test successfully.

--
Groetjes
joostje
47d3fcfe28f2a83497e79d9bc7d5087c-4a1ee1fddab4648175518cb4c1c9edb3ed0e89f0
--
To UNSUBSCRIBE, email to [EMAIL PROTECTED] 
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]

Re: Serious compiler problem

[Joost Witteveen]

Sebastian Lövdahl wrote:
Hello! I have a very big problem with GCC (maybe this is wrong thread, 
but maybe it isn't GCC related at all). When I try to compile a specific 
file, GCC gives me this:

In file included from globals.h:11,
from main.c:19:
/usr/lib/gcc-lib/i486-linux/3.3.4/include/stddef.h:151: error: syntax 
error before "typedef"
Well, if you have a simple file like:
$ cat > f.c
int h
#include 
Then trying to compile the file f.c, you'll get the error message you have.
Conclusion: somewhere before you include stddef.h, you have some syntax 
error (possibly a missing ";" as in the file f.c above).

--
Groetjes
joostje
47d3fcfe28f2a83497e79d9bc7d5087c-4a1ee1fddab4648175518cb4c1c9edb3ed0e89f0
--
To UNSUBSCRIBE, email to [EMAIL PROTECTED] 
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]

Re: permissions problems in ssh session as root (WAS: apt-get install and upgrade errors)

[Joost Witteveen]

james derry wrote:
#hi, bela,
#thanks for your reply. as a sanity check, i made sure that as root to 
run `apt-get upgrade` from root directory (/). same problem.

#looking further, it seems the problem may have to do with root 
permissions problems, and not with apt-get or dpkg at all. logged in as 
root on ssh, i cannot `touch` a test file to /usr/bin/ directory:

touch: creating '/usr/bin/testFile': Permission denied
The only thing I can imagine is that /usr (or /usr/bin) is 
network-mounted from some other host, that doesn't permit writing.
What does /proc/mounts say about how /usr is mounted?

(If /usr were mounted read-only on a local device, root also would not 
be allowed to touch there, but the error message would be different).


--
Groetjes
joostje
47d3fcfe28f2a83497e79d9bc7d5087c-4a1ee1fddab4648175518cb4c1c9edb3ed0e89f0
--
To UNSUBSCRIBE, email to [EMAIL PROTECTED] 
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]

Re: fetchmail stopped working (woody).

[Joost Witteveen]

s. keeling wrote:
fetchmail: timeout after 300 seconds waiting for listener to respond.
fetchmail: client/server synchronization error while delivering to SMTP host 
mail.spots.ab.ca
fetchmail: 5.9.11 querying mail.spots.ab.ca (protocol POP3) at Wed 03 Nov 2004 
03:26:56 PM MST: poll completed

poll mail.spots.ab.ca with proto POP3
   user 'keeling' there with password 'MY_PASSWORD' is 'keeling' here options 
fetchall
Seems you try to deliver to your (local) SMTP server.
I've had problems with that too (SMTP server refusing to accept the
email for example for SPAM protection reasons)
To solve that, I'm using in .fetchmailrc (delivers to procmail):
poll pop3.sever.tld protocol pop3 username joostje password PASSWD
  mda "/usr/bin/procmail  -d  %T"
--
Groetjes
joostje
1f1c8cdd94ddb3264368a4ec8c8369f4-592b9df7b0a20369a50b47f8db73f8b635bc637c
--
To UNSUBSCRIBE, email to [EMAIL PROTECTED] 
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]

Re: number of files in directory?

[Joost Witteveen]

Thomas Adam wrote:
 --- Lance Hoffmeyer <[EMAIL PROTECTED]> wrote: 

How can I get a count of the number of files in a directory?
directory + subdirectoies?

Crudely:
ls -1 | wc -l
(note the "-1" option to 'ls' is hyphen-one, NOT lower-case L, which is
what the option to 'wc' is).
Note, too, that the "-1" option is not deeded, as ls detects that stdout
isn't a terminal, and it wil use one-collumn output with or without the 
"-1".

--
Groetjes
joostje
1f1c8cdd94ddb3264368a4ec8c8369f4-592b9df7b0a20369a50b47f8db73f8b635bc637c
--
To UNSUBSCRIBE, email to [EMAIL PROTECTED] 
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]

Re: bash_logout & xsession help needed

[Joost Witteveen]

Michael Graham wrote:
I have this simple .xsession
#!/bin/bash -l
gnome-session
and this even simpler .bash_logout
echo hello > /tmp/logout
When I logout from a console the file /tmp/logout is created (as expected)
but when I logout of gnome the file isn't created. Can anyone explain this
behaviour?
Wouldn't a .xsession file like this do this:
$ cat ~/.xsession
gnome-session
echo hello > /tmp/logout
Or am I missing something?
--
Groetjes
joostje
1f1c8cdd94ddb3264368a4ec8c8369f4-592b9df7b0a20369a50b47f8db73f8b635bc637c
--
To UNSUBSCRIBE, email to [EMAIL PROTECTED] 
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]

Re: Ctrl+U in Firefox location bar

[Joost Witteveen]

[KS] wrote:
Hello all,
I have noticed this behaviour in Firefox recently. When the cursor is in 
the location bar and I press "Ctrl+U" to delete the text, Firefox opens 
the "View Source" window. This is annoying as it changes a basic linux 
shortcut's functionality.
Well, that depends on what you think is `basic linux shortcut 
functionality'. But yes, I'm used to using the emacs-like shortcuts, and 
it annoyed me too.

To fix (in firefox and all/many? other gnome apps), eighter run
gnome-keybinding-properties from the shell and set
`text editing shortcuts' to "Emacs", or do the same in
Applications->DesctopPreferences->KeyboardShortcuts from the Gnome menu.
--
Groetjes
joostje
1f1c8cdd94ddb3264368a4ec8c8369f4-592b9df7b0a20369a50b47f8db73f8b635bc637c
--
To UNSUBSCRIBE, email to [EMAIL PROTECTED] 
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]

Re: grub, raid1 help

[Joost Witteveen]

Richard Weil wrote:

One that doesn't work is:
title   Debian GNU/Linux, kernel 2.6.7 DISK1
root(hd0,0)
kernel  /boot/vmlinuz-2.6.7 root=/dev/md0 ro
savedefault
boot
Strange. I have:
title  hda: Linux 2.4.23-raid
root   (hd0,4)
kernel /boot/kernelo-2.4.23-raid root=/dev/md0
title  hdc: Linux 2.4.23-raid
root   (hd1,4)
kernel /boot/kernelo-2.4.23-raid root=/dev/md0
and it works OK.
If you get kernel panics after GRUB has loaded your kernel,
I would suppose the kernel just isn't RAID1 enabled.
--
Groetjes
joostje
8b534037349343140df39156acbede5c-de402678bcb40866dc0b29def3543aaa06193eef
--
To UNSUBSCRIBE, email to [EMAIL PROTECTED] 
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]

Re: Subversion version 1.1

[Joost Witteveen]

Alan Chandler wrote:
I tried to find out why subversion 1.1 is still sitting in experimental rather 
than making its way into unstable, but I was unable to find out any reason, 
despite seaching development and this mailing list for any discussion about 
it.

Anyone any ideas when we will see it.
 

See:
http://lists.debian.org/debian-devel/2004/10/msg01520.html
(and the few replies to it).
--
Groetjes
joostje
8b534037349343140df39156acbede5c-de402678bcb40866dc0b29def3543aaa06193eef
--
To UNSUBSCRIBE, email to [EMAIL PROTECTED] 
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]

Re: Thunderbird 0.8 and Filters

[Joost Witteveen]

Joost Witteveen wrote:
Ed Sutherland wrote:
Alexander Sack wrote:
Ed Sutherland wrote:
What's happened to filters in mozilla-thunderbird 0.8? The menu

I have no problems running filters etc.. Anyone having filter 
problems too, please attach more info to: bug #272954.

The problem is that when Thunderbird 0.8 included the one local 
folder choice, they disabled the ability to create filters based on a 
message. (You can still create a filter by hand and it will work, but 
they ability to create filters based on a selected message no longer 
works.)

The workaround is to create another account, don't choose the global 
inbox and then the create filter based on message works. This is like 
the time the mozilla folks dropped postscript printing support fo 
xprint, angering a huge portion of the users. I hope the filter 
feature returns in the next release.

OK.
I just tested yesterdays nightly (30 sept), and it still had the feature
(not able to create filter from message, able to click "run filters on 
folder", but doesn't work).

I just downloaded the image from
ftp://ftp.mozilla.org/pub/mozilla.org/thunderbird/nightly/2004-10-27-05-0.9/thunderbird-i686-linux-gtk2+xft.tar.gz
and the `feature' has been removed. I can choose `create filter from 
message', and "run filters on folder" seems to work too now.

--
Groetjes
joostje
09f04f72db6c6b12477b6202b91bf26e-6bed28733ed1c289493a8f1f7eb6ed3eb71909e6
--
To UNSUBSCRIBE, email to [EMAIL PROTECTED] 
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]

Re: Lynx doesn't dump to stdout?

[Joost Witteveen]

Roozemond, D.A. wrote:
Hi Matthijs,
 

Example:
lynx -dump
http://www.ticketmaster.nl/html/searchResult.htmI?keyword=carlton&l=NL
| grep resultaten

No need to understand all the above - If you change the '&' in the
webpage address to '\&', it's working:
Although I usually prefer to enclose the argument with ''-signs (easier,
and I don't have to search for all "&", "$", " " etc signs):
lynx -dump 
'http://www.ticketmaster.nl/html/searchResult.htmI?keyword=carlto\&l=NL' 
| grep resultaten

--
To UNSUBSCRIBE, email to [EMAIL PROTECTED] 
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]

Re: anyone tried chroot_safe?

[Joost Witteveen]

Joey Hess wrote:
Greg Norris wrote:
Has anyone here has tried out chroot_safe[1]?
I haven't read any of the code, but based on their documentation, so
long as you trust the binary you're chrooting, it should be as safe as
regular chroot. The paranoid part of me suspects that a malicious binary
could run under chroot_safe and manage to avoid running chrooted,
although it might have to find an exploit a hole in chroot_safe to do
so.
Looking at the code, it seems as though chroot_safe simply uses the
normal chroot() call; I would think the binary running would not be
able to see the difference between `real' chroot and chroot_safe
(and should thus not be able to exploit bugs in chroot_safe).
But as long as you trust the binary program you're chrooting, and
are only concerned about its behavior when fed untrusted data or the
like, after being chrooted, this seems like a perfectly safe and rather
handy way to go about chrooting it.
Oh and also, there's no reason a simple program without LD_PRELOAD magic
couldn't automatically set up a chroot environment for a program to run
in.
Indeed.
chroot_safe looks interesting, but I fail to see why they choose the
name "chroot_safe". I see nothing in the code (or the description)
that makes it safer than real chroot. It does make it a lot easier
to set up a chroot environment, that is what LD_PRELOAD is used for:
with the chroot_safe.so library in LD_PRELOAD, you don't need to
fill /bin, /lib/* etc with everything you normaly need to get chroot
running.
I think a better name would have been "chroot_simple", or "chroot_easy".
--
To UNSUBSCRIBE, email to [EMAIL PROTECTED] 
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]

Re: New user Q: Best way to stay up to date on "testing"?

[Joost Witteveen]

Jacob S wrote:
On Fri, 8 Oct 2004 23:13:15 +0200
"Dan Roozemond" <[EMAIL PROTECTED]> wrote:
 

While we're at it - suppose someone is the only administrator of a
debian(stable) system connected to the internet permanently, with SSH,
Postfix and Bind exposed to the 'big bad' world. Say that someone is
lucky enough to take a vacation, and is not able to connect to the
machine for two weeks. How dangerous is it to have 'apt-get update;
apt-get upgrade' ran automatically every day?
No, this scenario is not entirely hypothetically ;)
   

Well, let's just say that I wouldn't do it unless I were going to be
looking for a new job while enjoying that 2 weeks of vacation. :-)
With Stable it should be rare for it to be a problem, for Sarge it
shouldn't be a problem very often, but there is still a chance for
problems in there somewhere.
 

That's how it should be. However, IIRC, all (most?) security bugs in 
packages
like ssh, bind, etc, were present both in testing and in stable.


--
To UNSUBSCRIBE, email to [EMAIL PROTECTED] 
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]

Re: [Way OT] Re: GMAIL Invites..!

[Joost Witteveen]

Carl Fûrstenberg wrote:
On Thu, 07 Oct 2004 12:40:52 +0100, robin <[EMAIL PROTECTED]> wrote:
cr wrote: 
On Wed, 06 Oct 2004 11:57, Cybe R. Wizard wrote: 
On Tue, 05 Oct 2004 08:24:10 -0500 Rodney Richison
<[EMAIL PROTECTED]> wrote: ... 
I run a list for techs/consultants. (ChannelVar.com) They are a VERY
qualified bunch of guys. (Admittedly not linux users though) :) The

have anyone a copy of the site so we can check what it meens?
[...]
http://web.archive.org/web/*/ChannelVar.com

--
To UNSUBSCRIBE, email to [EMAIL PROTECTED] 
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]

Re: Grub won't timeout

[Joost Witteveen]

Shaul Karl wrote:
On Fri, Oct 08, 2004 at 12:00:38AM -0600, Jules Dubois wrote:
(I would have sword GRUB started numbering at 1, not 0, but perhaps I'm
mistaken.

  I believe you are wrong. It starts at 0. This should be documented.
Please check the documentation to see who is right.
GRUB religiously starts numbering at 0, througout the programme.
(hd0,0) is the first partition on the first harddisk;
default 0 means the first entry in the menu.lst file.
See also the grub info file, "Configuration":
 # By default, boot the first entry.
 default 0
   The first entry (here, counting starts with number zero, not one!)
will be the default choice.
--
Thanks,
joostje
--
To UNSUBSCRIBE, email to [EMAIL PROTECTED] 
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]

Re: sources.list entries?

[Joost Witteveen]

robin wrote:
Carl Fink wrote:
On Sat, Oct 02, 2004 at 03:27:13PM -0400, Skip Evans wrote:
 

Could someone send to me, or the list, entries from their own 
sources.list file that I can use to access a wider range of 
repositories?
  

Instructions:
1. Load www.debian.org.
2. Click on Search.
3. Search for sources.list
4. Click on the first result.
Good advice:
1. Learn to ask smart questions from Eric Raymond at
  
--  Carl Fink [EMAIL PROTECTED]
Jabootu's Minister of Proofreading
http://www.jabootu.com
 

Though the search page on debian.org has only just come on-line, last 
day or two, before which it was disabled for sometime
However, google has been online somewhat longer, and a google search for
site:www.debian.org sources.list
works too.
--
To UNSUBSCRIBE, email to [EMAIL PROTECTED] 
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]

Re: make xconfig & xfree???

[Joost Witteveen]

Carl Fink wrote:
On Thu, Sep 30, 2004 at 09:03:05AM -0600, Justin Guerin wrote:

I would recommend using some of the other suggestions given in this thread 
before using xhost +.

On my box, I tend to issue xhost local:+, which is (I would think)
pretty safe.
If you are sure you're the only person with login access to your host, yes.
Anyone else though with an account on your host will be able to see 
everything you type, though, if you ever give out a login to someone else.

Thanks,
joostje
--
To UNSUBSCRIBE, email to [EMAIL PROTECTED] 
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]

Re: make xconfig & xfree???

[Joost Witteveen]

Jeremy Brown wrote:
Robert Tilley wrote:
Make was issued as root.  Why can't root connect to the xserver?
[EMAIL PROTECTED]:/usr/src/linux-2.6.8.1# make xconfig
 HOSTCC  scripts/basic/fixdep
 HOSTCC  scripts/basic/split-include
 HOSTCC  scripts/basic/docproc
 HOSTCC  scripts/kconfig/conf.o
 HOSTCC  scripts/kconfig/kconfig_load.o
 HOSTCC  scripts/kconfig/mconf.o
 HOSTCXX scripts/kconfig/qconf.o
 HOSTCC  -fPIC scripts/kconfig/zconf.tab.o
 HOSTLLD -shared scripts/kconfig/libkconfig.so
 HOSTLD  scripts/kconfig/qconf
scripts/kconfig/qconf arch/i386/Kconfig
qconf: cannot connect to X server
make[1]: *** [xconfig] Error 1
make: *** [xconfig] Error 2
Any ideas welcome, Bob
 

Might want to try doing:
bash$ xhost +
as a regular user, before attempting to connect to the X server as root.
It's a lot safer to put in root's .bashrc:
if test "$LOGNAME" != "root"; then
  export XAUTHORITY=/home/$LOGNAME/.Xauthority
fi
This will give user root access to the X session of the logged-in-user.
What you suggested will give _every_ user access to the X session,
including the ability to read passwords etc.
(And, if the server accepts TCP connections, "xhost +" allows everyone 
on the internet to connect, and look while you type your passords etc).

--
joostje
--
To UNSUBSCRIBE, email to [EMAIL PROTECTED] 
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]

Re: Firefox Install

[Joost Witteveen]

On Wed, 29 Sep 2004 14:14:12 -0500, Jacob S <[EMAIL PROTECTED]> wrote:

As root, run firefox-installer. This will be easiest if you are logged
into X as root.

IMNSHO this is not wise, doing this will only create havoc behind the
back of your package manager (dpkg), the only place you can safely
install something without using dpkg is in your home or in
/usr/local... =)
The firefox-installer asks the user where to install firefox.
So just tell it to install in /usr/local, and you should be fine

--
To UNSUBSCRIBE, email to [EMAIL PROTECTED] 
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]

Re: LaTeX with Emacs

[Joost Witteveen]

Ali Nassar wrote:
Hi All,
I am using emacs on Debian-Linux to edit LaTex files. Is there any tool in
emacs to make a delimiter check? I mean to count the number of left
delimiters in the file and compare it with the number of right delimiters
and then give me an error message if a delimiter is missing.
Is chktex what you are looking for?
Long ago I remember using a tool to check (La)TeX files, and it worked 
quite well. Don't know what the name was though. chktex came out of 
google, and it's a GNU program, and part of Debian.

--
To UNSUBSCRIBE, email to [EMAIL PROTECTED] 
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]

Re: keeping woody

[Joost Witteveen]

Max Moritz Sievers wrote:
On Tuesday 28 September 2004 21:01, Clive Menzies wrote:
I don't think there is any non-free stuff on them. Time to google to verify
that. 

Take a look at the Virtual Richard M. Stallman:
http://packages.debian.org/stable/admin/vrms
Brilliant!
However,
# dpkg --purge unrar
dpkg - warning: ignoring request to remove unrar which isn't installed.
# vrms
Non-free packages with status other than installed on foko
unrar ( pur)
  1 non-free packages, 0.1% of 1587 installed packages.
I mean, I like being vokal about the badness of non-free software,
but does vRMS really have to complain about a non-free package I
once installed, but have no longer installed?
From my /var/lib/dpkg/status:
Package: unrar
Status: purge ok not-installed
Priority: optional
Section: non-free/utils
Architecture: i386
Anyway, thanks for the suggestion about vrms
--
To UNSUBSCRIBE, email to [EMAIL PROTECTED] 
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]

Re: keeping woody

[Joost Witteveen]

Clive Menzies wrote:
# this doesn't:
deb http://security.debian.org/debian-security woody/updates main contrib non-free

Ah!  Well this does here:
deb http://security.debian.org/ woody/updates main 
Reminds me of the time my box was rooted. Must have been 1997 or 
thereabouts, and I was constantly keeping my box uptodate with
the latest unstable (testing didn't exist, nor did apt), but I 
(religiously) updated only the really free stuff -- I didn't want to 
pollute my system with anything else.

I forgot ofcource that I did once install a pop3 server from the uni of 
washington (that wasn't free back then, but I believe I couldn't find a 
free alternative). And that server had a bug in it, and it hadn't been 
updated for months.

Well, all that just to mention that only using the security updates from 
main may be dangerous, if you already have non-free packages installed.

Anyway http://www.debian.org/security  advertises
deb http://security.debian.org/ woody/updates main contrib non-free
so I suppose that should work too. (without the debian-security part)
--
Groetjes,
joostje
--
To UNSUBSCRIBE, email to [EMAIL PROTECTED] 
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]

Re: Editing PDF forms

[Joost Witteveen]

Richard Lyons wrote:
On Monday 27 September 2004 21:01, Hugo Vanwoerkom wrote:
 

Dave Howorth wrote:
   

Does anybody know of a way to edit PDF forms then save the result?
 


I fill out the IRS PDF forms and save them.
Use pstoedit and transfig to create a .fig file of the PDF, like:
   

What is a .fig file?
 

A .fig file is what is created by the command you choose to cut out:
$ pstoedit -f "fig:-startdepth 999"  %d.fig

--
To UNSUBSCRIBE, email to [EMAIL PROTECTED] 
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]

Re: 2.6.8-3 kernel freezes on Inspiron 8100 at isapnp

[Joost Witteveen]

Andrea Vettorello wrote:
On Fri, 24 Sep 2004 21:51:09 +0800, Uwe Dippel <[EMAIL PROTECTED]> wrote:
 

(today I'm kind of unlucky with the daily update, it seems: Thunderbird
broke, XMMS broke and now:)
After the update from 2.6.8-2 (??), but 2.6.8 in any case 2.6.8; the
kernel freezes at isapnp. I tried three times (Power-Off); and then booted
to 2.6.3; the install-kernel; without any problem.
I wonder if I'm the only one and should file a bug-report ?
   

I've experienced a freeze plugging/unplugging my usb printer with 2.6.8-2...
 

There is actually a bug in 2.6.8.1 that causes an OOPS when I switch 
off  my USB printer
in 2.6.8.1 -- a bug that apparently already was there in 2.6.5 (or before).

Here's the patch that worked for me (and now apparently has been applied 
by Greg KH to
the kernel):

Below is the email with the patch in it:
Vojtech:
This bug was reported back in July, and I sent out a patch but apparently 
it never got to you.  The usblp driver was calling usb_buffer_free() from 
usblp_cleanup(), which runs after disconnect() if a user process holds the 
device open.  But once the usb_device is gone usb_buffer_free() will 
oops.  The patch frees the buffers in usb_disconnect() instead.

Recently Joost Witteveen reported the same oops and found that the
patch solved it for him.  So there shouldn't be problems with accepting 
it.

Alan Stern

Signed-off-by: Alan Stern <[EMAIL PROTECTED]>
= drivers/usb/class/usblp.c 1.108 vs edited =
--- 1.108/drivers/usb/class/usblp.c 2004-09-09 05:24:11 -04:00
+++ edited/drivers/usb/class/usblp.c2004-09-20 14:33:03 -04:00
@@ -397,10 +397,6 @@
{
info("usblp%d: removed", usblp->minor);
-   usb_buffer_free (usblp->dev, USBLP_BUF_SIZE,
-   usblp->writebuf, usblp->writeurb->transfer_dma);
-   usb_buffer_free (usblp->dev, USBLP_BUF_SIZE,
-   usblp->readbuf, usblp->readurb->transfer_dma);
kfree (usblp->device_id_string);
kfree (usblp->statusbuf);
usb_free_urb(usblp->writeurb);
@@ -1159,6 +1155,10 @@
usb_set_intfdata (intf, NULL);
usblp_unlink_urbs(usblp);
+   usb_buffer_free (usblp->dev, USBLP_BUF_SIZE,
+   usblp->writebuf, usblp->writeurb->transfer_dma);
+   usb_buffer_free (usblp->dev, USBLP_BUF_SIZE,
+   usblp->readbuf, usblp->readurb->transfer_dma);
if (!usblp->used)
usblp_cleanup (usblp);


Andrea
 


--
To UNSUBSCRIBE, email to [EMAIL PROTECTED] 
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]

Re: Need help finding NIC driver

[Joost Witteveen]

On 2004-03-29, [EMAIL PROTECTED] <[EMAIL PROTECTED]> wrote:
> Sorry for the delay in response.
>
> Yes, I tried that, as well as insmod tulip (no idea, figured it was
> worth trying).
>
> System detects "eth0", which I fear is my disabled (in BIOS) onboard
> ethernet and not my NIC card, which I ironically added when SuSE was
> not able to use my onboard ethernet.  Before you get excited, it has
> since stopped functioning altogether.  I get an error message when I
> try the PPPoE configuration utility, which says it cannot access the
> Access Compressor or something along those lines.  I tried with
> firewall up and down.
>
> Does anyone know of a link where I can find a driver?  This is not a
> rare NIC card, it's not brand new nor too old.

http://www.kernel.org has drivers for (almost?) all common NIC cards,
that is, Linux itself knows how to handle them and you don't need to
download a 'driver' from a different web-page.

> I guess I could go
> buy another one, but would prefer not to.
>
> Would a UNIX driver work for it?  They do have a UNIX driver on the
> 3COM website.

If it's a 3com card, why not try to insmod the various 3com modules?
If you've got debian kernel & modules installed, you can figure out
the available modules using:

find /lib/modules/`uname -r`/ -name '*3c*

and insmod the appropriate one.


Groetjes,
joostje


-- 
To UNSUBSCRIBE, email to [EMAIL PROTECTED] 
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]

Re: 2nd NIC: Netgear FA-330, insmod problem

[Joost Witteveen]

On 2004-03-29, Max <[EMAIL PROTECTED]> wrote:
> Hi,
>
> I am trying to put in a second NIC, a Netgear FA-330/FA-331.
>
> lspci shows:
> 00:09.0 Ethernet controller: 3Com Corporation 3c905C-TX
> 00:0c.0 Ethernet controller: National Semiconductor Corporation DP83815
> (MacPhyter)
>
> The first one is my old card, so I guess the second one is the new card. I
> am using the 2.4.18-k6 kernel, which didn't seem to come with a ns83815
> driver, but has a ns83820 driver. I tried using that, but get insmod "No
> such device" errors.

from linux-2.4.24/Documentation/Configure.help:
National Semiconductor DP8381x series PCI Ethernet support
CONFIG_NATSEMI
  This driver is for the National Semiconductor DP83810 series,
  which is used in cards from PureData, NetGear, Linksys
  and others, including the 83815 chip.
  More specific information and updates are available from
  .

And from linux-2.4.24/drivers/net/Makefile:
 obj-$(CONFIG_NATSEMI) += natsemi.o

So, I suppose the module you're looking for is called
natsemi.o. Is that one available?


Groetjes,
joostje


-- 
To UNSUBSCRIBE, email to [EMAIL PROTECTED] 
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]

Re: mkraid could not lock sda2

[Joost Witteveen]

On 2004-03-24, Justin Guerin <[EMAIL PROTECTED]> wrote:

> Note that there's no reason to raid your swap space.
> The kernel does this for you.

Depends on what RAID you want to use for swapspace.
If it's RAID1 (as mentioned in a previous post), then
obviously the kernel doesn't do that for you.

-- 
groetjes,
joostje



-- 
To UNSUBSCRIBE, email to [EMAIL PROTECTED] 
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]

Re: problems with automount

[Joost Witteveen]

On 2004-03-23, Karsten Bolding <[EMAIL PROTECTED]> wrote:
> Hello
>
> I'm having som problems with the automounter...
>
> /etc/auto.master:
> -
> /mnt/debian /etc/auto.debian
>
> /etc/auto.debian:
> -
> debian  -ro gate:/public/ftp/debian

[...]
>
> If I do:
> ls /mnt/debian
> or
> cd /mnt/debian
>
> I don't get anything mounted - only the empty directory. I can do the
> mounting manually without any problems.

Shouldn't you do:
cd /mnt/debian/debian
as /etc/auto.debian specifies "debian" as subdirectory of the /mnt/debian
dir in /etc/auto.master ?

Groetjes,
joostje


-- 
To UNSUBSCRIBE, email to [EMAIL PROTECTED] 
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]

Re: ..lwresd, was meant to be: BIND 9: rcnd: connection refused

[Joost Witteveen]

In article <[EMAIL PROTECTED]>, Arnt Karlsen wrote:

>> > When I start Bind from '/etc/init.d/bind9 start', everything looks
>> > fine, like nothing wrong. But I found out that Bind acctually not
>> > working properly. /etc/init.d/bind9 reload/restart will give this
>> > error message:
>> > 
>> > Stopping domain name service: named
>> > rndc: connect failed: connection refused
> 
> ..I've seen these too.  ;-)
>   
>> The magic is in /etc/bind/rndc.conf (and the corresponding key
>> in /etc/bind/named.conf).
>> 
>> Create the magic using rndc-confgen.
>> 
>> In my case, lwresd was somehow installed, messing up the
>> communication. 
> 
> ..how???

How was lwresd installed? I don't know. Maybe from first install, maybe
it was pulled in with some other package.

How did lwresd mess up the communication between rndc and named(bind9)?
Apparently, lwresd was listening on port 953, so that named couldn't
bind on that port (and thus rndc could not communicate to named).

>> Removing lwresd (and kill-ing the process that stayed on even after I
>> removed the package) solved it.
> 
> ..in my case, this took 3 minutes and 9 to 25 seconds

What took 3 minutes?
>, lwresd on or off
> made at best a 2 second difference _if_ it mattered, and I'm not sure of
> that.

I guess it depends on who was started first, lwresd or named.
If named is started first, it will listen on port 953, and then
everything should work OK.

As I don't know why I would want to have lwresd, I just removed it.
I should file a bugreport against one of the packages, but don't
do that now, as I don't have time to investigate what package is
wrong etc.

-- 
joostje


-- 
To UNSUBSCRIBE, email to [EMAIL PROTECTED] 
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]

Re: DINB 9: rcnd: connection refused

[Joost Witteveen]

In article <[EMAIL PROTECTED]>, Phillipus Gunawan wrote:
> Hi there,
> 
> I am setting up a BIND for my local network (anydomain.bom) acessible
> only for local domain. This is the second time I build BIND, my old
> Debian crashed after I accidently unplug one of my hdd-ide cable (there
> are 3 hdd(s) ) and Debian kernel starting panic...
> 
> When I start Bind from '/etc/init.d/bind9 start', everything looks fine,
> like nothing wrong. But I found out that Bind acctually not working
> properly. /etc/init.d/bind9 reload/restart will give this error message:
> 
> Stopping domain name service: named
> rndc: connect failed: connection refused

The magic is in /etc/bind/rndc.conf (and the corresponding key
in /etc/bind/named.conf).

Create the magic using rndc-confgen.

In my case, lwresd was somehow installed, and messing up the
communication. Removing lwresd (and kill-ing the process that
stayed on even after I removed the package) solved it.

Oh, and after you've setup rndc.conf and named.conf, you need
to kill named once manually, so that it re-reads the named.conf
information.


--
joostje


-- 
To UNSUBSCRIBE, email to [EMAIL PROTECTED] 
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]

Re: Failure mounting extra partition on boot

[Joost Witteveen]

In article <[EMAIL PROTECTED]>, Bruce wrote:
>  /dev/hda3 / ext3 defaults,noatime 1 1
>  /dev/hda1 /mnt/hda1 ext3 defaults,auto,users,exec,noatime,notail 0 0
[..]
>  [EMAIL PROTECTED] mount /mnt/hda1
>  mount: wrong fs type, bad option, bad superblock on /dev/hda1,
>or too many mounted file systems
>  [EMAIL PROTECTED]
> 
> However, it will mount properly if I give the following command
> 
>  [EMAIL PROTECTED] mount /dev/hda1 /mnt/hda1

The only difference I see is that in this case, you don't specify
the filesystem type, and mount figures it out itself. So, are
you sure it's ext3? Does
  mount -t ext3 /dev/hda1 /mnt/hda1
work? what does dmesg (or /var/log/kern.log) report about the filesystem
type after you successfully mounted it?


joostje


-- 
To UNSUBSCRIBE, email to [EMAIL PROTECTED] 
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]

Re: FUBARed my install, wont boot :(

[Joost Witteveen]

In article <[EMAIL PROTECTED]>, Andreas Janssen wrote:
> Hello
> 
> Kent West (<[EMAIL PROTECTED]>) wrote:
> 
>> I suspect you need to add the line
>>   initrd=/initrd.img
>> to /etc/lilo.conf and rerun lilo. (I would think you could get in by
>> feeding the initrd parameter to the bootup process, but you may need
>> to get in first to figure out what parameter syntax to use.)
> 
> Can passing an initrd parameter to the /kernel/ work? If the kernel does
> not have support for the root or boot file system compiled in, and the
> initrd is located on the root or boot partition, how can the kernel
> access the initrd on its own? If I understand this concept correctly,
> then the bootloader must know about the initrd. Does anyone know
> details on whether this can work with lilo or grub?

>From the GRUB info file (see point 3):

  1. Set GRUB's root device to the same drive as GNU/Linux's. Probably
 the command `find /vmlinuz' or similar can help you (*note find::).

  2. Load the kernel:

  grub> kernel /vmlinuz root=/dev/hda1

 If you need to specify some kernel parameters, just append them to
 the command. For example, to set `vga' to `ext', do this:

  grub> kernel /vmlinuz root=/dev/hda1 vga=ext

 See the documentation in the Linux source tree for the complete
 information on the available options.

  3. If you use an initrd, execute the command `initrd' (*note
 initrd::) after `kernel':

  grub> initrd /initrd

  4. Finally, run the command `boot' (*note boot::).


So, GRUB loads the initrd in memory. the initrd node of the GRUB info 
file notes:

initrd
--

 - Command: initrd file ...
 Load an initial ramdisk for a Linux format boot image and set the
 appropriate parameters in the Linux setup area in memory. See also
 *Note GNU/Linux::.



joostje


-- 
To UNSUBSCRIBE, email to [EMAIL PROTECTED] 
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]

Re: Sid et DHCP

[Joost Witteveen]

In article <[EMAIL PROTECTED]>, Matthias Hentges wrote:
> Am Di, den 10.02.2004 schrieb [EMAIL PROTECTED] um 10:49:
>> Bonjour tout le monde,
>> 
>> je viens de faire un dist-upgrade sur ma woody (noyau 2.6.1) pour passer en 
>> sid. Tout s'est bien passé jusqu'au reboot.
>> En effet, maintenant je n'ai plus accès au réseau (via dhcp)! alors qu'avant 
>> cela ne posait pas de problème. Est-ce une erreur courante? J'ai absolument 
>> besoin d'aide, c'est très important pour moi de résoudre ce problème le plus 
>> rapidement possible
> 
> [...]
> 
> You do realize that the majority on this planet does not speak french,
> huh?

I dont think English is spoken by 50% of this planet's inhabitants.
Maybe we should all switch to mandarin?

-- 
joostje


-- 
To UNSUBSCRIBE, email to [EMAIL PROTECTED] 
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]

Re: Permissions

[Joost Witteveen]

In article <[EMAIL PROTECTED]>, Johann Koenig wrote:
> On Monday February  9 at 07:45pm
> Joshua Jankowski <[EMAIL PROTECTED]> wrote:
> 
>> As I have been quite intelligent in setting permissions on my debian
>> server, I am here to see if anyone has a solution.  In my attempt to
>> write recursive permissions on one of my directories, I hit enter a
>> little too prematurely with / as the designated folder.  Quickly
>> noticing the error, I hit ctrl-c to stop the operation but as you can
>> guiess, it was not soon enough.  It overwrote the permissions that
>> were set by debian in the /bin folder and unknown others.
>> 
>> Is there a utility or way to easily(or not) fix the default
>> permissions?
> 
> I b0rked my old server in a similar fashion. Only sane way to fix it is
> a clean install. I tried checking the permissions on a similar computer,
> but there are so many files with special ownership/permissions that I
> very quickly got very frustrated.
> 
> Re-install is the best way. (This is the only scenario I've run into
> that really *required* a reinstall. Everything else I've been able to
> fix.)

Well, if it's only the /bin directory, I'd simply chmod -R 0755 /bin
and hope for the best (su, ping, login should be setuidroot, mount
and unmount are as well on my system, but I never use that).

If a particular package/command gives problems, you can always use
  apt-get --reinstall install package
to fix the file permissions of that particular package.

All permissions of all debian-installed files (except /etc) are
described in the .deb files, so if you really wanted to, you should
be able to write a little script that gets all installed .debs and
corrects the permissions.

-- 
joostje


-- 
To UNSUBSCRIBE, email to [EMAIL PROTECTED] 
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]

Re: md5 failed inside 3.0r2 ISOs

[Joost Witteveen]

In article <[EMAIL PROTECTED]>, AIRCOMP, S.L. wrote:
> De: Adam Aube <[EMAIL PROTECTED]>
> Para: [EMAIL PROTECTED] <[EMAIL PROTECTED]>
> Fecha: lunes 9 de febrero de 2004 17:00
> Asunto: Re: md5 failed inside 3.0r2 ISOs
> 
> 
>>On Monday 09 February 2004 07:35 am, AIRCOMP, S.L. wrote:
>>> I just downloaded the 7 ISO images. The MD5 check of the images is fine
>>> but when I mount it -o loop and do a second check using /md5sum.txt
>>> hundreds files fail or are zero size.
>>
>>So the MD5 on the images themselves is fine, but when you check individual
>>files in the image it fails? What if you copy files out to your main
>>filesystem and check them there?
>>
>>I have seen behavior like this with CDs - if I try to verify the MD5 hash
>>generated when the file was on the hard drive, it doesn't match. If I
>>copy the file from the CD to my hard drive, the hashes match again.
>>
>>Adam
> 
> 
> Excuse me but the behavior you suggest is even more weird than what I'm
> experiencing. Any file should give equal MD5 hashes no matter if it is on a
> disk or on a CD-ROM.
> 
> This happens with at least two ISO images (1-NONUS and 2) downloaded from
> the .es mirror (ftp.es.debian.org/debian-cd/3.0_r2/i386/). If this is
> abnormal (and I believe it is) then there's something *very* wrong at that
> mirror.

If the MD5 sum of the total .iso is correct than one can be fairly[1]
sure the mirror is perfectly OK


[1] OK, maybe the .es guys broke the MD5 checksum mechanism, and intentionally
put different data in the .iso. But if they really did break MD5,
why didn't they also fix-up the internal md5 checksums?

joostje


-- 
To UNSUBSCRIBE, email to [EMAIL PROTECTED] 
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]

Re: 'apt-get dist-upgrade' not working.

[Joost Witteveen]

In article <[EMAIL PROTECTED]>, Adam Bogacki wrote:
> Hi, I'm running unstable and have not been able to do a successful
> 'apt-get dist-upgrade --fix-missing' for 2-3 weeks now.  That is a big 
> backlog in unstable.
> 
> The key message seems to be
> 
>> E: Unable to parse package file /var/lib/dpkg/status (1)
> 
> 'apt-get update' gives me the message
> 
>> W: Couldn't stat source package list http://mirror.pacific.net.au 
>> unstable/main Packages 
>> (/var/lib/apt/lists/mirror.pacific.net.au_debian_dists_unstable_main_binary-i386_Packages)
>>  
>> - stat (2 No such file or directory)
>> E: Unable to parse package file /var/lib/dpkg/status (1)
>> E: The package lists or status file could not be parsed or opened.

If te package file /var/lib/dpkg/status really is messed up, 
not much is going to work. Do you still have a file in
/var/lib/dpkg/status-old ?
If so, make a copy of it in some safe place (and also of
any messed-up /var/lib/dpkg/status file), and then maybe try

cp /var/lib/dpkg/status-old /var/lib/dpkg/status

and run again. I don't know what you can do if you really don't
have a status file any more.


-- 
joostje


-- 
To UNSUBSCRIBE, email to [EMAIL PROTECTED] 
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]

Re: restrict users from seeing each other

[Joost Witteveen]

In article <[EMAIL PROTECTED]>, Adam Aube wrote:
> On Friday 06 February 2004 05:24 pm, Kevin Coyner wrote:
>> Also, separate, but related question:  If a user has access via ssh, is
>> there any way to restrict him/her to just their /home directory?
> 
> I have heard (but not tried) that specifying the home directoy in /etc/
> passwd like this:
> 
> /home/username/.
> 
> will accomplish this. Again, I haven't tried this, but it might be worth a 
> shot.

I've just tried it, and it doesn't work.
Wouldn't know why it would work, anyway.

(debian-testing).

joostje


-- 
To UNSUBSCRIBE, email to [EMAIL PROTECTED] 
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]

Re: Phone dialer for Linux

[Joost Witteveen]

In article <[EMAIL PROTECTED]>, Darik Horn wrote:
> > I'm looking for something that can be used in a way sort of like the
> > following:
> >
> > ~$ somedialer  -P{ulse}  -D /dev/ttys0 -N <1 234 567 8910>
> 
> Try minicom, tip, and cu.  Use whatever best fits your taste.
> 
> 
> > I've searched and searched and I just can't seem to find a simple,
> > command-line phone dialer for use with pulse/rotarty (ie non-touch
> > tone) phone lines.
> 
> Set pulse dialing in the modem initialization string.  In your terminal 
> emulator options, change all instances of "ATDT" to "ATDP".

Yes, but some (most?) tone-modems ignore the "ATDP", and dail pulse
anyway. I found that out when using my 56k modem in Bulgaria
(I bought it in the Netherlands). Absolutely no way to make it dail
using pulses, I had to go and buy a modem in a local shop.

joostje


-- 
To UNSUBSCRIBE, email to [EMAIL PROTECTED] 
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]

Re: HOWTO - Speed up IDE HD's - raid

[Joost Witteveen]

In article <[EMAIL PROTECTED]>, Alvin Oga wrote:
> 
>> RAID 1 is one strategy for getting 'faster' read going.
> 
> 
> http://www6.tomshardware.com/storage/2329/fastrak66-14.html
> 
> 
> "its" linearly faster for reading ... but also linearly slower for writing
>:-)

First, I cannot see that statement (writing linearly slower) confirmed
on the page you mention, and second, I wonder why that would be.
When writing one block of data to a raid1 array of N disks, the block
indeed needs to be written N times -- but it can be sent to each disk
simultaniously, so it should take constant time.

Sure, writing to a raid 1 array will never be as fast as straight to
disk, but it shouldn't be all that much slower.

Thanks,
joostje


-- 
To UNSUBSCRIBE, email to [EMAIL PROTECTED] 
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]

Re: dvips TeX printer driver

[Joost Witteveen]

In article <[EMAIL PROTECTED]>, Haines Brown wrote:
>> 
>> On Fri, Feb 06, 2004 at 11:37:00AM -0500, Haines Brown wrote:
>> > When I migrated to debian, I lost my ability to print LaTeX files. No
>> > information is sent to the printer.
>> > 
>> > The printer driver is dvips, and I assumed that it was included in the
>> > LaTeX package, for that was the case with my LaTeX installation under
>> > RedHat. Under Debian, is the driver a separate package, and if so what
>> > is it called? 
>> > 
>> I would guess that what is happening is that you are running dvips and
>> expecting seing your file coming from the printer. Well, at least in
>> sid, when you run dvips it works as if you were using 
>> "dvips -o file.ps file.dvi", so a solution could be to turn this off.
> 
> Interesting. I always used dvips to create the .ps file and print it
> simultaneously, and you are correct that it works under woody as if
> the -o option were present. But the man offered no clue as to how to
> "turn off" the option which is not on to begin with (unless it is
> present in a dvips script somewhere).

It does:

dvips -o '!lpr' file.dvi

But indeed the change of default behaviour did happen silently,
and could be better described in the manual page, at the thop of
the man file, it still reads

- The program dvips takes a DVI file file[.dvi] produced by TeX (or by
- some other processor such as GFtoDVI) and converts it to PostScript,
- normally  sending  the  result directly  to the (laser)printer 

-- 
joostje


-- 
To UNSUBSCRIBE, email to [EMAIL PROTECTED] 
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]

Re: dvips TeX printer driver

[Joost Witteveen]

In article <[EMAIL PROTECTED]>, Haines Brown wrote:
> When I migrated to debian, I lost my ability to print LaTeX files. No
> information is sent to the printer.
> 
> The printer driver is dvips, and I assumed that it was included in the
> LaTeX package, for that was the case with my LaTeX installation under
> RedHat. Under Debian, is the driver a separate package, and if so what
> is it called? 

I'm using tetex, and it is wonderfully wel. dvips is in the tetex-bin
package.

I wonder why you say "no information is sent to the printer". Do you
have dvips installed, but is it simply not working? Can you grenerate
proper .ps files, and can you see them with ghostscript (or gv, etc).

Thanks,
joostje


-- 
To UNSUBSCRIBE, email to [EMAIL PROTECTED] 
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]

Re: apt-gey: I can't find anywhere this answer! :(

[Joost Witteveen]

In article <[EMAIL PROTECTED]>, 82roby wrote:
> hi all, this is my first question :)
> 
> please help me answering this question:
> 
> which packages did I install with apt-get on the 1st of february?

Don't think you can get a real list. You can do however

cd /var/lib/dpkg/
ls -altr *.list|less

You'll see the packages list, sorted by the time of last installation
of each package. However, if you've installed package X on feb 1,
and updated it again on feb 2, it will come up as feb 2 only.

The other option is looking at /var/lib/dpkg/status and status-old,
but that doesn't go back that far.



-- 
To UNSUBSCRIBE, email to [EMAIL PROTECTED] 
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]

Re: XFree 4.2 server DIY (chrooted env)

[joost witteveen]

Ah, forgot to mention:


> 5 On the main system, startup xfs (apt-get install xfs, remove
>   the `no-listen = tcp' line in /etc/X11/fs/config),
>   and install xdm and activate it 
>   (easiest way is to add a "*" line in /etc/X11/xdm/Xaccess),
>   and maybe remove those pesky `no tcp' lines from xdm too.

And, add the local fontserver to the XF86Config-4 file in the
chroot dir. For example, the "Files" section could look like this:

Section "Files"
FontPath"tcp/127.0.0.1:7100"
EndSection


-- 
To UNSUBSCRIBE, email to [EMAIL PROTECTED] 
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]

XFree 4.2 server DIY (chrooted env)

[joost witteveen]

Hi,

I recently bought a MSI MS-6368 mainboard, and although Xfree 4.1 does
support the Trident Microsystems CyberBlade/i1, it is somewhat flaky
(see below). So, I wanted to try out Xfree 4.2.

I did not want to install XFree 4.2 the way XFree suggests
(mv /usr/X11R6 backup; install new in /usr/X11R6), as I was
afraid Debian wouldn't like it (Debian also installs it stuff in /usr/X11R6,
so things may brake).

So, I decided to try to install XFree 4.2 in a little chroot-ed environment,
where the installation will not harm my (new!) debian system.
I wanted to _only_ install the server part of XFree 4.2, all the
client binaries I'll use from my Debian system (they work fine, anyway).
So, I want the XFree 4.2 server to contact my Debian 4.1 XDM daemon.

As there seem to be some people who want Debian to have 4.2, I decided
to post to the list what I did (as at least one step wasn't fully obvious),
I decided to post what I did to the list.

Well, in words:

1 Get Xxserv.tgz, Xmod.tgz, Xlib.tgz (the latter only needed if you want
  anything other than a black/white system) from 
  ftp://ftp.xfree86.org/pub/XFree86/4.2.0/binaries/Linux-ix86-glibc22/
  (depending on your system etc)

2 create a directory, with copies of your /etc, /dev, /bin, /lib
  root-directories (simlinks won't do, I'm sure you can leave out
  most of the files in the dirs, but I didn't want to find out what).

3 create usr/X11R6, and unpack the X*.tgz files in here, downloaded in
  step 1
  make a symlink usr/bin/X111 -> /usr/X11R6/bin

4 **
  create tmp (mkdir tmp; chmod a+rwx tmp; chmod o+t tmp)
  and in your ROOT dir, do
  mv /tmp /tmp-old
  ln -s $YOUR_XFREE_CHROOT_DIR/tmp /tmp

5 On the main system, startup xfs (apt-get install xfs, remove
  the `no-listen = tcp' line in /etc/X11/fs/config),
  and install xdm and activate it 
  (easiest way is to add a "*" line in /etc/X11/xdm/Xaccess),
  and maybe remove those pesky `no tcp' lines from xdm too.

6 chroot $YOUR_XFREE_CHROOT_DIR
  X -query 127.0.0.1
  Ready.


Or, for those who understand sh better (step 5 above you still need to do)


#!/bin/bash

set -e
set -x
cd /home/joostje/rommel

rm -rf Xtest
mkdir Xtest
cd Xtest

mkdir -p usr/X11R6
mkdir usr/bin
ln -s /usr/X11R6/bin usr/bin/X11
mkdir -p var/log

for f in dev bin lib etc; do
  echo Copying $f
  cp -a /$f . || true
done

mkdir tmp
chmod a+rwx tmp
chmod o+t tmp

for f in Xxserv.tgz Xmod.tgz Xlib.tgz; do
  echo unpacking $f
  #cp ../Xfree4.2.0/$f .
  wget ftp://ftp.xfree86.org/pub/XFree86/4.2.0/binaries/Linux-ix86-glibc22/$f
  zcat $f | (cd usr/X11R6; tar -xf -)
done

#Here, I copy a modified XF86Config file:
#cp /home/joostje/rommel/Xfree4.2.0/etc/X11/XF86Config-4 etc/X11

mv /tmp /tmp-tmp
ln -s `pwd`/tmp /tmp


(and now, `chroot Xtest`, and `X -query 127.0.0.1`)




Anyway, more on why I needed Xfree4.2 for my  MSI MS-6368 
Trident Microsystems CyberBlade/i1:

- I couldn't go back to VC mode with 4.1. Whenever I pressed
  Alt_Cntr F1, the screen would go into power-save mode,
  and X wouldn't revive from it (well, I could go back to VC1,
  and restart X from there, but my old X session died).

- Every now and then, the mouse pointer would be drawn about 50
  pixels to the right of where the click-action would be. Very
  annoying, and even restarting X wouldn't work. I had to actually
  reboot the computer to get rid of it.

- XFree 4.1 didn't like lines like
   Option  "SWCursor" "true"
   Option  "NoAccel" "true"
  in the "DEVICE" Secion in XF86Config-4. Whenever I met one of them,
  X wouldn't start, screen would go into power-save mode.

  XFree 4.2 did like those lines, and, using the "SWCursor" option,
  my mouse-action now is where the mouse pointer actually
  shows. Very convenient!

Also, it seems XFree 4.2 likes higher video resolutions. I don't know
why, but it automatically elected a much higer resolution, from
an identical XF86Config-4 file.


Hoping to be of some use to someone out there, and hoping
not to start a flame-ware (it's about xfree 4.2 after all),

joostje


-- 
To UNSUBSCRIBE, email to [EMAIL PROTECTED] 
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]

Re: galeon 0.10.6 Packages anywhere ?

[joost witteveen]

> download the 0.10.6 source code .tar.gz and do the following:
> 
> tar -zxvf galeon-0.10.6.tar.gz
> cd galeon-0.10.6
> ./configure

Done that, installed tons of .dev (and other) packages, but
whatever I do, it will not get past the ./configure
stage.

The stage where things go wrong is:

checking for gnome-config... /usr/bin/gnome-config
checking for GNOME - version >= 1.2.0... yes
checking for additional GNOME modules...  vfs 
checking for oaf-config... /usr/bin/oaf-config
checking for OAF - version >= 0.6.2... no

When I run oaf, I get:

$ /usr/bin/oaf-config --version

So the configure script is obviousli wrong.

Looking at the conftest.c program the ./configure script generated,
it is obviously wrong:
it declares:
  int major, minor, micro;
but later uses:
  if ((liboaf_major_version != 0) ||
  (liboaf_minor_version != 6) ||
  (liboaf_micro_version != 5))

So it looks to mi like an error in the ./configure script 
(that script has nearly the same .c file in it)

$ md5sum ./configure ../galeon-0.10.6.tar.gz
8fa8cc7ca45b3d090029834a449ffbe1  ./configure
eda0f7dc904298dcdff09e871825eaf7  ../galeon-0.10.6.tar.gz

Am I going mad?



Below is the more complete ./configure log (from after it tests for
the oaf version):

checking for GNOME - version >= 1.2.0... yes
checking for additional GNOME modules...  vfs 
checking for oaf-config... /usr/bin/oaf-config
checking for OAF - version >= 0.6.2... no
*** Could not run OAF test program, checking why...
*** The test program failed to compile or link. See the file config.log for the
*** exact error that occured. This usually means OAF was incorrectly installed
*** or that you have moved OAF since it was installed. In the latter case, you
*** may want to edit the oaf-config script: /usr/bin/oaf-config
checking for gconf-config... /usr/bin/gconf-config
checking for GCONF - version >= 0.12... yes
checking for libglade-config... no
checking for libglade... no
checking for orbit-config... /usr/bin/orbit-config
checking for ORBit - version >= 0.5.1... yes
checking for orbit-idl... /usr/bin/orbit-idl
checking for gdk-pixbuf-config... no
checking for GDK_PIXBUF - version >= 0.9.0... no
*** The gdk-pixbuf-config script installed by GDK_PIXBUF could not be found
*** If GDK_PIXBUF was installed in PREFIX, make sure PREFIX/bin is in
*** your path, or set the GDK_PIXBUF_CONFIG environment variable to the
*** full path to gdk-pixbuf-config.
checking for PR_Init in -lnspr4... yes
checking for PL_strcmp in -lplc4... yes
checking for PL_NewHashTable in -lplds4... yes
checking for xpcom... yes
checking for gdk_superwin_get_type in -lgtksuperwin... no
checking for gtk_moz_embed_get_type in -lgtkembedmoz... no
checking for Mozilla... no
configure: error: 
*** Mozilla 0.9 is required 
*** A package for 0.9 is available here:
*** http://people.redhat.com/blizzard/software/.




$ dpkg -l mozil*
Desired=Unknown/Install/Remove/Purge/Hold
| Status=Not/Installed/Config-files/Unpacked/Failed-config/Half-installed
|/ Err?=(none)/Hold/Reinst-required/X=both-problems (Status,Err: uppercase=bad)
||/ Name   VersionDescription
+++-==-==-
ii  mozilla0.9-0.0.3  An Open Source WWW browser for X and GTK+
ii  mozilla-dev0.9-0.0.3  Header files for Mozilla development
un  mozilla-dmotif  (no description available)
rc  mozilla-smotif 5.0b1-2Mozilla Navigator

Re: /etc/rc* and locate

[joost witteveen]

> > > I have no
> > > program which runs update nightly.
> > 
> > I think you do. see /etc/cron.daily/find
> >
> Sorry, I didn't know that. Makes the anomaly even more difficult to
> understand though

Yes.

It looks like your updatedb doesn't do what it should do.
You could try and remove the file /var/lib/locate/locatedb,
and then run updatedb. Does it re-create the file? If so,
does locate still show the /etc/rc* files?

Does  /etc/updatedb.conf have any funky contents?

-- 
joostje

Re: announce galeon-0.10.6 unofficial package

[joost witteveen]

> > http://christophe.barbe.online.fr/debian/

> What apt says:

I use wget, and the installed galeon works!

Thanks christophe barbe!

-- 
joostje

Re: fvwm2

[joost witteveen]

Je 2001/05/23(3)/11:05, Joerg Johannes montris sian geniecon skribante:
> Hi List
> 
> Using sid in the most recent version with XFree4.0.3, I just switched to
> the good old fvwm I used to have in SuSE times (I never liked KDE and
> Gnome).
> 
> 1.) .Xresources. Well, the problem I have is: when starting emacs from
> the fvwm-mennu, it starts up with default values (white background,
> standard geometry,...). When I start it from an xterm, it starts with
> the options I set in .Xdefaults and .Xresources (In fact, these two are
> hardlinked). So, why doesn't fvwm read these files?
> 
> 2.) I use gdm as login manager (because of the "shutdown" option). When
> logging out of fvwm2, I don't get the login screen back, but only an
> empty root window with mouse pointer. Falling back to the login screen
> works with all other WM'S (e.g. enlightenment, gnome, kde). What could
> cause this?
> 
> 3.1) the menus (general). I have poked around in the reades, and read a
> lot about "hook files". I did not quite understand how to use them. I
> understood it in so far, that it is not very smart to have a personal
> menu in my .fvwm directory, because the system-wide menu file is changed
> by the menu-generation script. I can add entrys to 
> the sytem-wide menus by adding them to the several hook-files, but I
> cannot remove entrys, because I had to set up a personal menu and would
> lose the comfort of a script-updated menu when installing new programs.
> Now, what can I do to have custom menus?

To add entries in your personal menu tree, take a look at
any file in /usr/lib/menu/*, and copy/change a file to your ~/.menu dir

To remove the menu entries you don't want, (as in /usr/doc/menu/menu.txt.gz):

5.2. Specifying a ``no-menu entry''
---

 If a user wants to remove an entry from the system menu (in
 `/etc/menu'), then this will do the trick:

echo -n  > ~/.menu/package

 The zero-size file will tell `update-menus' that the corresponding
 package should not have any menu entries listed.


-- 
joostje

Re: Request

[joost witteveen]

Je 2001/05/16(3)/ 4:05, Martin Popovici montris sian geniecon skribante:
> Sir,
> 
>We are a computer sellers company from Romania and we are interested 
> 
> at this moment to buy ASAP 100 processors Intel Pentium III @ 933 MHz not 
> 
> boxed(tray) and 400 memories @128 Mb PC 133. 
> We are interested to get from your company your prices list, including 
> 
> any other products and 
> your payments and shipping  methods.

Complete distribution, including source: $0
shipping: via internet.

(CD's also available, at slightly higher cost).

-- 
joostje

Re: Copying one hard drive to another

[joost witteveen]

Je 2001/05/14(1)/11:05, Nathan E Norman montris sian geniecon skribante:
> On Sat, May 12, 2001 at 09:06:17PM -0600, Cameron Matheson wrote:
> > Hey,
> > 
> > I need to copy everything from hard drive to another.  I'm assuming this is
> > done w/ dd, but I need to be sure.  Could someone give me an example of how 
> > to
> > copy /dev/hda1 to /dev/hdb1?
> 
> mount /dev/hdb1 /mnt
> cd 
> find -xdev | cpio -padm /mnt

that (well, I'd do it with tar) is probably the best.
BUT -- I've got 3 hardisks of 20 Gig, and when I copy
the contents from one to another, I do use dd
(dd if=/dev/hda of=/dev/hdc bs=102400), and it works
miraculausly well:). I do suppose it will fail if the
disks aren't exactly the same size, though. And I've
never tried copying partitions -- but, it they are
the same sizes, it might work to.


-- 
joostje

X Window, Right Ctrl -> Control?

[joost witteveen]

Hello all,

Could anyone tell me how to make the right Ctrl key work
(just as the left one) as Control? At the moment it seems
to do just nothing (although xev does notice it).

I've tried adding 
  RightCtlControl
to my /etc/X11/XF86Control file, but that doesn't seem to work.
(I'm using XFree 3.1.6 now, but will switch to 4.0 if that's what
it takes).

Thanks!

joostje

Re: update-menus not working? Here is my info.

[joost witteveen]

At 2001/02/10(6)/10:02, Xucaen wrote:

> > > ?package(jed):needs="text" \
> > > section="Apps/Editors" \
> > > title="jed" \
> > > command="/usr/bin/jed"
[ ... and `dpkg -l jed' shows it's installed OK, still updates-menu ignores jed]

A strange story! I just copied your `jed' file to my own
/etc/menu, and changed the `?pacakge(jed)' to `?package(local.jed)',
and it works OK for me (I don't have jed installed, so I just
added the `local.')

> [EMAIL PROTECTED]:/etc/menu$ ls -l
> total 20
> -rw-r--r--1 root root 397 Jan 22  2000 README
> -rw-r--r--1 root root 127 Jan 25 12:07 aim
> -rw-r--r--1 root root  97 Jan 30 08:12 jed
> -rw-r--r--1 root root 116 Jan 30 07:30 mc
> -rw-r--r--1 root root 103 Jan 30 07:30 xjed

That's fine too.
> 
> I tried update-menus -d.
> There is too much output to capture

Well, you can try
 update-menus -d  2> file
and you can later read file slowly. (using `less' for example)
(and, a UNIX tip: see `man script' if you want a more
general solution to `too much output to capture')

> , but from
> what I am able to see it never processes
> etc/menu/jed or /etc/menu/xjed (but how can that
> be?)

There realy should be a line like:

Update-menus[1659]: Reading menuentryfile /etc/menu/jed

in the `update-menus -d' output. Are you certain even
that is missing? Then something strange is going on.

Oh, maybe you already told me, but what version of menu
are you using? (dpkg -l menu)

Thanks,
joostje

Re: update-menus not working?

[Joost Witteveen]

On Tue, Jan 30, 2001 at 10:12:25AM -0800, Xucaen wrote:
> is /etc/menu/jed:
> 
> ?package(jed):needs="text" \
> section="Apps/Editors" \
> title="jed" \
> command="/usr/bin/jed"

Looks fine to me. Do you hava a `jed' package installed
(does `dpkg -l jed' give a line that starts with `i'?). If not,
then you should use something like `?package(local.jed):...'.

Also, I assume the file-permissions are OK (not executable, and
readable).

> when I run update-menus (as root) it doesn't add

You can try `update-menus -v' or `update-menus -d', to see more
output. (with -d, IIRC, you should see /etc/menu/jed come by).

If that doesn't work, please email me. Maybe it is a bug in the menu
package.

Thanks,
joostje

GDB: Cannot access memory at address 0x40017df0

[Joost Witteveen]

Hi,
this morning, I did the upgrade of bind, and because of that, my
libc6 was upgraded. After that, I've been unable to run gdb
on my favorite little program. Does this soudn familiar to
anyone? (ie, the error: `Cannot access memory at address 0x40017df0').

After that, I upgraded my whole system (apt-get upgrade), so now
all libraries are up-to-date (as wel as the .dev packages), but
the error message stays.

(Yes, I can put a breakpoint at the first line of the program, and
still I get the same error message).

$ gdb q
GNU gdb 5.0
Copyright 2000 Free Software Foundation, Inc.
GDB is free software, covered by the GNU General Public License, and you
are
welcome to change it and/or distribute copies of it under certain
conditions.
Type "show copying" to see the conditions.
There is absolutely no warranty for GDB.  Type "show warranty" for
details.
This GDB was configured as "i686-pc-linux-gnu"...
(gdb) run
Starting program: /home/users/r+d/central2/joost/config/q 
Cannot access memory at address 0x40017df0

$ ldd q
libcrypt.so.1 => /lib/libcrypt.so.1 (0x40021000)
libresolv.so.2 => /lib/libresolv.so.2 (0x4004e000)
libldap.so.2 => /usr/lib/libldap.so.2 (0x4005f000)
liblber.so.2 => /usr/lib/liblber.so.2 (0x40085000)
libc.so.6 => /lib/libc.so.6 (0x4008f000)
libnsl.so.1 => /lib/libnsl.so.1 (0x4019b000)
libsasl.so.7 => /usr/lib/libsasl.so.7 (0x401b)
/lib/ld-linux.so.2 => /lib/ld-linux.so.2 (0x4000)
libdb2.so.2 => /usr/lib/libdb2.so.2 (0x401bc000)
libdl.so.2 => /lib/libdl.so.2 (0x4020)
libpam.so.0 => /lib/libpam.so.0 (0x40203000)


Thanks,
joostje

.*? still greedy (perl,python)?

[joost witteveen]

Hi,
I noticed that .*? in perl and phython, if used at the beginning
of an unanchored (without ^) regular expression still behaves half-greedy:

perl:
  "hello" =~ /(.*?)l/;
  print "\$1 = $1\n";

python:
  import re
  r=re.search("(.*?)l","hello")
  print "$1=",r.group(1)

In both cases, "he" is printed, while the real minimal match obviously
would result in an emtpy $1 or r.group(1). This is the behavour
if I put the (.*?) at the end (ex.: /h(.*?)/)

Is this a bug in perl/python itself, or maybe in the documentation?

Or am I misunderstanding something?

Should I file bugs?

(notice that, when replacing .*? with .*, both perl and python find
"hel", so it seems the greedyness to the right-hand-side is controled
OK by the ?, but not to the left-hand-side).

(Yes, I know I can `fix' it by adding .* to the beginning of the reg-expr).

$ dpkg -l python\* perl\*|grep ^i
ii  python-base1.5.1-3An interactive object-oriented scripting lan
ii  python-doc 1.5.2p1-4  Documentation for the scripting language Pyt
ii  python-gtk 0.6.5-1GTK support module for Python.
ii  python-net 1.5.1-3TCP/IP and various Internet support modules 
ii  python-tk  1.5.1-3Tk support module for Python (Tkinter).
ii  perl-5.004 5.004.05-6 Larry Wall's Practical Extracting and Report
ii  perl-5.004-bas 5.004.05-6 The Pathologically Eclectic Rubbish Lister
ii  perl-5.004-sui 5.004.05-6 Runs setuid Perl scripts.
ii  perl-5.005 5.005.03-7.1   Larry Wall's Practical Extracting and Report
ii  perl-5.005-bas 5.005.03-7.1   The Pathologically Eclectic Rubbish Lister
ii  perl-5.005-doc 5.005.03-6 Man pages and pod docs for Perl
ii  perl-base  5.004.05-1.1   Fake package assuring that one of the -base 

-- 
joostje

Modem pulse dailing?

[joost witteveen]

I want my modem to dail using pulses, not tones (some phonelines in
Bulgaria need it).

After setting `Pulse' in the pppconfig menu, chat seems to
use ATDP as dail string, but I still hear tones.

The Linux Newbie Guide (http://sunsite.auc.dk/linux-newbie/FAQ.htm)
claims that all Hayes-compatible modems should accept pulse dailing,
so what am I doing wrong?

(from /var/log/ppp.log:)

Aug 20 02:10:10 muso chat[1050]: send (ATDP8800999^M)

Thanks 
joostje

Alt Keys -> Meta (under X11)

[joost witteveen]

Does anyone know how to make the Alt keys generate Meta? 
That is I want Alt-F to generate Meta-F (under Emacs, bash, etc).


I've tried setting in my XF86Config:

LeftAlt Meta
RightAltMeta

But that didn't work.

I used to work with "XkbDisable" uncommented (in XF86Config), but
I believe that's depreciated in Debian, and anyway causes Backspace/Delete
problems in various programs. So I commented it out again, and hoped
to make the Alt keys generate Meta in some other way.


Thanks,
joostje

ppp/chat: BLACKLISTED after (many) tries

[joost witteveen]

After several (about 10) unsucessful tries (engaged), chat consistantly
refuses to rediail my ISP. In syslog, I get:

Jan 18 20:34:57 muso modprobe: can't locate module char-major-108
Jan 18 20:34:58 muso pppd[640]: pppd 2.3.9 started by root, uid 0
Jan 18 20:34:59 muso chat[642]: abort on (BUSY)
Jan 18 20:34:59 muso chat[642]: abort on (NO CARRIER)
Jan 18 20:34:59 muso chat[642]: abort on (VOICE)
Jan 18 20:34:59 muso chat[642]: abort on (NO DIALTONE)
Jan 18 20:34:59 muso chat[642]: abort on (NO ANSWER)
Jan 18 20:34:59 muso chat[642]: send (ATZ^M)
Jan 18 20:34:59 muso chat[642]: expect (OK)
Jan 18 20:35:01 muso chat[642]: ATZ^M^M
Jan 18 20:35:01 muso chat[642]: OK
Jan 18 20:35:01 muso chat[642]:  -- got it 
Jan 18 20:35:01 muso chat[642]: send (ATDT08000230517^M)
Jan 18 20:35:01 muso chat[642]: expect (CONNECT)
Jan 18 20:35:01 muso chat[642]: ^M
Jan 18 20:35:04 muso chat[642]: ATDT08000230517^M^M
Jan 18 20:35:04 muso chat[642]: BLACKLISTED^M
Jan 18 20:35:46 muso chat[642]: alarm
Jan 18 20:35:46 muso chat[642]: Failed
Jan 18 20:35:46 muso pppd[640]: Connect script failed
Jan 18 20:35:47 muso pppd[640]: Exit.

Does anyone know how to make chat stop thinking that provider
is BLAKCLISTED? Yes, I should get a real ISP, but that's not
possible in this country (all the others make me pay telephone
chageres).

(Another silly thing is that, once any phone number is blacklisted,
every phone number after that is blakclisted too. So, to change
ISP, I actually have to reboot!)

Thanks,
-- 
joostje

presentation graphics

[joost witteveen]

> I want to produce some simple bar charts showing the difference between
> different products.  Gnuplot doesn't seem to be able to do what I want
(just
> nice clean bars with numbers at the tops and descriptions at the bottom).

> Is there a good program to do this easily?  Preferrably something I can
run
> from a makefile...

Well, I'm not claiming the stuff below is good:). But is is
easy, and you can generate it from a makefile (just make the
makefile insert your own table where I now got that postscript
array).


%!PS

/width 10 def

/r{(bar.ps) run} def

[
[10 (foo)]
[20 (bar)]
[30 (foobar)]
[29 (fooba)]
[50 (fbfb)]
[20 (bar)]
]

/bar {
aload pop
/title exch def
/y exch def
gsave
0.5 setgray
width 0 rlineto
0 y rlineto
width -1 mul 0 rlineto
closepath
fill
grestore
gsave
width 2 div y width 2 div add rmoveto 90 rotate title show
grestore
} def

/Times-Roman findfont 10 scalefont setfont
100 100 moveto
{
width 0 rmoveto
bar
} forall

showpage

__
FREE Email for ALL! Sign up at http://www.mail.com

presentation graphics

[joost witteveen]

> I want to produce some simple bar charts showing the difference between
> different products.  Gnuplot doesn't seem to be able to do what I want
(just
> nice clean bars with numbers at the tops and descriptions at the bottom).

> Is there a good program to do this easily?  Preferrably something I can
run
> from a makefile...

Well, I'm not claiming the stuff below is good:). But is is
easy, and you can generate it from a makefile (just make the
makefile insert your own table where I now got that postscript
array).


%!PS

/width 10 def

/r{(bar.ps) run} def

[
[10 (foo)]
[20 (bar)]
[30 (foobar)]
[29 (fooba)]
[50 (fbfb)]
[20 (bar)]
]

/bar {
aload pop
/title exch def
/y exch def
gsave
0.5 setgray
width 0 rlineto
0 y rlineto
width -1 mul 0 rlineto
closepath
fill
grestore
gsave
width 2 div y width 2 div add rmoveto 90 rotate title show
grestore
} def

/Times-Roman findfont 10 scalefont setfont
100 100 moveto
{
width 0 rmoveto
bar
} forall

showpage

__
FREE Email for ALL! Sign up at http://www.mail.com

Re: X, middle mouse,Artec, do see output in /dev/mouse

[Joost Witteveen]

On Sun, Nov 21, 1999 at 11:01:30AM +0100, Colin Marquardt wrote:
> * joost witteveen <[EMAIL PROTECTED]> writes:
> 
> > I've two mice here who's middle mouse button I cannot get to work
> > under X. However, cat-ing /dev/mouse shows that i do get output:
> [...]
> > So, it seems like it should be possible for X to detect
> > midle mouse presses.
> 
> Does xev recognise it?

Thanks for the idea, but no, it doesn't :(. So, I assume
I don't have to try this:
> 
> Maybe you need to play with xmodmap (search xmodmap with / pointer,
> *not* / mouse).  E.g. xmodmap -e "pointer = 1 2 3"

> Or... does gpmconfig recognise the mouse as a slightly different
> model which gives the middle button by default (which is then
> maybe/hopefully also known to X).


But I've already tried all possible X mice. And everything works
with the Microsoft setting, except for the middle button. Strange
thing is that I've now got two such mice, bought at different times,
and from different suppliers.

Thanks,

Re: procmail

[joost witteveen]

Je 1999/11/20(6)/18:11, GECOS montris sian geniecon skribante:
> Hi,
> I am trying to filter my mail with procmail.
> I gues in debian you need a forward file.
> Every time I try to use the .forward file all my mail disappears.
> 
> I will attach jmy .forward file with this message.
> It is called rick right now since I do not want it to work.
> 
> -- 
> Rick Knebel
> [EMAIL PROTECTED]

> "|IFS=' ' && exec /usr/local/bin/procmail -f- || exit 75 #rknebel"

If you are using debian's procmail, the binary is in /usr/bin/procmail,
not /usr/local/bin.

For the rest it seems like it should work, although I wonder
why you don't just use the example from the procmail manual:

"|exec /usr/bin/procmail"

Or, if using exim:

|/usr/bin/procmail


-- 
joostje

X, middle mouse,Artec, do see output in /dev/mouse

[joost witteveen]

Hi,

I've two mice here who's middle mouse button I cannot get to work
under X. However, cat-ing /dev/mouse shows that i do get output:


left button:
muso:/dev# cat  mouse |od -x
000 0060 4000  0060 4000  0060 4000

Middle button:
muso:/dev# cat  mouse |od -x
000 0040 4000  0040 4000  0040 4000

right button:
muso:/dev# cat  mouse |od -x
000 0050 4000  0050 4000  0050 4000

So, it seems like it should be possible for X to detect
midle mouse presses.

For a mouse that does have a working middle button under X, I get,
in the same order:
000 0060 4000  0060 4000  0060 4000
000 0040 2000 0040  0040 2000 0040 
000 0050 4000  0050 4000  0050 4000

So, that is slightly different. Does anyone know how to make the
mouse work?
(I tried about every possible Pointer setting in XF86Config, but nothing
works).

ii  xserver-common  3.3.3.1-10 files and utilities common to all X servers
ii  xserver-s3  3.3.3.1-10 X server for S3 chipset-based graphics cards

Thanks,
-- 
joostje

Re: cpp macros: quotes and spaces (# and ##)

[joost witteveen]

Je 1999/08/27(5)/19:08, joost witteveen montris sian geniecon skribante:
> I've got a define
> 
> #define WRAP_STAT __xstat
> 
> and from the macro WRAP_STAT, I want to define
> 
> #define NEXT_STAT next___xstat
>  and
> #define WRAP_STAT_QUOTE "__xstat"

[What doesn't work:]
> #define NEXT_STAT next_ ## WRAP_STAT
[..]
> #define QUOTE(a) #a
> #define WRAP_STAT_QUOTE QUOTE(WRAP_STAT)
[..]

OK, if everything else fails, read the docs. So, now I finally found
(in cpp.info, `Argument Prescan') how to do it. Still, I'd very much
like to hear if anyone knows how portable this is:

#define WRAP_STAT __xstat

#define NEXT_STAT(a,b,c) (CONCAT_LEVEL(WRAP_STAT))(a,b,c)
#define CONCAT_LEVEL(X)CONCAT(X)
#define CONCAT(X) next_ ## X
NEXT_STAT(1,2,3)

#define WRAP_STAT_QUOTE QUOTE_LEVEL(WRAP_STAT)
#define QUOTE_LEVEL(x) QUOTE_MACRO(x)
#define QUOTE_MACRO(y)  #y
WRAP_STAT_QUOTE

(at least for GNU cpp it works!)

> (Please also reply to me in person, as I am not subscribed to this
> list).

-- 
joostje

cpp macros: quotes and spaces (# and ##)

[joost witteveen]

I've got a define

#define WRAP_STAT __xstat

and from the macro WRAP_STAT, I want to define

#define NEXT_STAT next___xstat
 and
#define WRAP_STAT_QUOTE "__xstat"

To define NEXT_STAT in terms of WRAP_STAT, I tried
#define NEXT_STAT next_ ## WRAP_STAT
(with the ## to get rid of the space in between next_ and __xstat).
But this doesn't work, as apparently cpp takes everything after ##
literally, and now NEXT_STAT is defined ad next_WRAP_STAT.
Does anyone know how to remove the space, and make cpp expand
the macro WRAP_STAT?

The other, WRAP_STAT_QUOTE, also seems difficult. I tried
#define QUOTE(a) #a
#define WRAP_STAT_QUOTE QUOTE(WRAP_STAT)
but then again, cpp fails to expand the WRAP_STAT, resulting
in effectively
#define WRAP_STAT "WRAP_STAT"
not exactly what I wanted.
Does anyone have a way to fix that?


(Please also reply to me in person, as I am not subscribed to this
list).

joostje

debian-user@lists.debian.org

[joost witteveen]

If I put in my configure.in:

AC_DEFINE_UNQUOTED(foo(a), bar a)

then in the created config.h (or on the gcc commandline) I get:

#define foo (a) bar a
   ^space

This doesn't work, I need to have:
#define foo(a) bar a

At the moment I `solve' this by doing:
  
  sed -e 's/ *(/(/' < config.h > sed-config.h

at the end of the configure.in file. But I'm sure there is a better
way to solve this?



If there isn't, maybe another way out would be to get rid of
the macro arguments altogether. But then I got the following
problem:

I want to define

#define WRAP_STAT  __xstat
#define NEXT_STAT(a,b,c) next___xstat(a,b,c)

To get rid of the macro arguments, I just defining WRAP_STAT by
configure, and then in some .h file defining NEXT_STAT in terms of
WRAP_STAT:

#define NEXT_STAT(a,b,c) next_ ## WRAP_STAT(a,b,c)

(with the ## to remove the space between next_ and __xstat). But
apparently cpp takes everything after the ## litaraly, and doens't
expand the macros any more, so that NEXT_STAT(1,2,3) now expands
to `next_WRAP_STAT( 1 , 2 , 3 )'.
So, does anyone know of a way to get rid of the space in between
next_ / WRAP_STAT(a,b,c)?


ii  autoconf2.13-7 automatic configure script builder
ii  automake1.4-2  A tool for generating GNU Standards-complian
ii  gcc 2.91.66-2  The GNU (EGCS) C compiler.
ii  cpp 2.91.66-2  The GNU (EGCS) C preprocessor.


(BTW, please also reply to me personally, as I'm not subscribed to
this list).
-- 
joostje

LILO on hdc (to be hda)

[joost witteveen]

So I thought I know how to work with LILO. And I *thought* done
this many times before (years ago). But...

I want to prepare a harddisk (that should later become the root FS)
So I put it in my system (it was found as hdc), and mounted it
as /mnt


I put in the lilo.conf file:

  boot=/dev/hdc
  root=/dev/hda1

Then I run lilo as

  lilo -r /mnt -C /etc/lilo.conf

and everything goes OK (appart from LILO warning about
/dev/hdc not being on the first disk). But then, when I
re-wire the box, and put the disk on the first IDE slot
(so that it becomes hda), the booting shows '01' ("Illegal command",
indicates non-supported disks). Could it be that LILO
is trying to read the kernel from hdc now (hdc now
is not just non-supported disk, it is a non-existing disk).

The problem basically is that I want to be able
to specify three things:
 - root filesystem after Linux has booted
 - the device to write boot table to
 - the device for lilo to use at booting to read the kernels etc.
And I only have `root=' and `boot=' arguments to specify all
three.

Am I something missing (I suppose so, but I cannot find it in
the docs)

(Please, CC me on replies, as I'm not subscribed to debian-users)

Thanks!
-- 
joostje

Re: UMAX 1220S, SCSI card (436P?)

[joost witteveen]

Je 1999/04/27(2)/ 9:04, H C Pumphrey montris sian geniecon skribante:

> On Mon, 26 Apr 1999, joost witteveen wrote:

> This sounds as if it could be the same card as is supplied with the Artec
> AT12 scanner. Here is the message I sent to SANE-Devel recently about
> this. I have updated it slightly since then. If you get this card to work,
> do please tell me!

Just tell mi where I can send that check for GBP 15 that you've just won!

It works all perfectly, even better than what you describe: I can
scan multiple images (well, I've scanned 3 now, but I assume it will
work indefinately).

All that I do differently is that I've got linux 2.2.3 here.
For the rest, I started my kernel with the same boot prompt as you
described, card got recognised, same for scanner, and sane just 
works. If I can help you by giving any more information, I'll be
glad to do so!

Many thanks,
joost

UMAX 1220S, SCSI card (436P?)

[joost witteveen]

As the SANE docs told me the UMAX 1220S should work OK,
I just bought one of those scanners.

The scanner came with it's own ISA SCSI adaptor. The vendor
already told me the card was somewhat strange, but well,
I went for it. Anyway, I cannot get Linux 2.2.3 to recognise
it (Simply turned on all SCSI interfaces).

So, does anyone know if the UMAX Astra 1220 S also works with
a normal SCSI card? (If so, I'll just buy one of those).
Or, maybe someone has tips how to get the card I've got to work?

Of cource, the stuff came without any information as to what
brand SCSI card I have. So, all I know about the card is that
the chip on it has written on it `DOMEX 436P, 9817, 002-D43P-001'

(Oh, please also CC me, as I'm not (yet) subscirbed to this list).

Thanks,
joost ([EMAIL PROTECTED])

Re: Backspace in xterm (again)

[Joost Witteveen]

skribis Gilbert Laycock:
> Pete> Does Alt-backspace work for anyone in an xterm (deleting the previous
> Pete> word on bash input, for example)?  This worked fine in Bo and before.
> 
> What kind of keyboard do you have? On my 104 key ("windows 95")
> keyboard, the "windows" key produces Meta and the Alt key produces Alt 
> when in X. Meta-backspace deletes the previous word in bash in an
> xterm. Alt-backspace just deletes one character. Using an rxvt it is
> the other way around. 
> 
> It is my understanding that Meta and Alt should be separate when there
> are enough keyboard buttons to allow it, and that Meta-backspace is
> the definitive way to delete a word, since it is emulating "emacs key
> bindings".

But then `Alt-Backspace' in emacs deletes one word, as it does on
the VC and so on. So I don't see the point in doing it different
on an xterm. Nor do I think it is good: When I press `Alt-Backspace',
I want something different to happen than when I just press `Backspace'.
Whatever `different' means is discutable, but just mapping the two keys
to the same action seems not quite the best thing.

Mapping Both Alt and Esc to the same thing seem s OK to me, as that's
what happens in most other environments too.

> 
> -- 
> 
>   Gilbert Laycock email:  [EMAIL PROTECTED]
>   Maths and Computer Science, http://www.mcs.le.ac.uk/~glaycock
>   Leicester Universityphone: (+44) 116 252 3902
> 

Re: Backspace in xterm (again)

[Joost Witteveen]

> > Does anybody know what I should do to get my beloved backspace back again
> > (yes, I want backspace to delete the char left to the curser, as it does
> > in the VC's).
> 
> Make sure the following is present in your /etc/X11/Xresources:

Thanks. But unfortunately, it _is_ already present!

> *Text.translations: #override ~Shift ~Meta Delete: 
> delete-next-character()
> XTerm*termName: xterm-debian
> *VT100*backarrowKey: false
> *VT100.Translations: #override BackSpace: string("\177")\n\
>Delete: string("\033[3~")\n\
>  Home: string("\033OH")\n\
>  End: string("\033OF")
> 
> I don't like introducing xterm-debian terminfo entry myself and just renamed
> it to xterm.

And my $TERM variable already is set to xterm-debian.

Backspace in xterm (again)

[Joost Witteveen]

Hi, here I am agian.

I upgraded my system yesterday (OK to slink, but it appears the same problem
is present in hamm), and now my bacspace generates an delete again.
(in an xterm that is. Emacs (X11) and VC are OK)

I already asked this question once, but I seem to have lost the responce
I got then. 

Does anybody know what I should do to get my beloved backspace back again
(yes, I want backspace to delete the char left to the curser, as it does
in the VC's).

Thanks,
[EMAIL PROTECTED] / [EMAIL PROTECTED]

Re: refocus photo?

[Joost Witteveen]

> On Fri, Aug 14, 1998 at 10:59:44PM +0200, Joost Witteveen wrote:
> > Hi,
> > 
> > I've got (at least) one photo taken while the camera was moving.
> 
> Only one? :)

Well, one that I care about. Much more, actually.

> 
> > I'd really like to get rid of the vageness caused by this motion.
> 
> Mmmh. I doubt that this is possible... but I'd like to be proofed wrong.
> Actually, missing sharpness is lost information.

Actaully, it isn't. If you take two pictures, and superimpose both
of them slightly (or more) displaced, it _is_ possible to excactly calculate
back what the original was (assuming you know what the displacement was). 

With a moved picuture it will probably be somewhat more difficult
(you don't know the displacement, and it's not two superimposed pictures,
but more like a continuas mess of them), but it should still be possible
(just a lot more work).

> How can this information be
> regained? 

OK, one simple 1D example (a 2D picture is just excactly the same)

assume the intensity of the `real' picture is

0022002222400

(with 0 begin black, 4 being white).

Now assume the camera was displaced one `step` halfway during the
exposure time:

00121012210001320
will be the result.

Now, with a few assumptions (the real difficulty is in the edges, but
assume one knows the outer two positions were black), it is quite
easy to calculate the image back:

The dot on position 3 of the `moved' image being 1 proves that
the real intensity was 2 (we know the first two positions
are black, because otherwise the `moved' image can never have two
initial black pixels, and so the `1' intensity was half of the
real intensity of position 3).

The `real' dot on position 3 begin intensity 2 means that half that
intensity was moved over to dot position 4. So of the intensity `2'
of that `moved pos 4' dot, 1 part is from dot 3, and 1 part must be
from dot 4. Thus also dot 4 must be intensity `2'. And so on.


> OTOH, maybe there are acceptable solutions for certain types of
> photos.
> 
> The only hint I can give you is reading the gimp manual. Maybe there are
> some tricks in there.

Only the `sharpen' stuff. What I want is a function that atually calculates
the displacement (not possible in the general case, but nearly
always possible in reality), and then inverts the `motion'



joost witteveen

refocus photo?

[Joost Witteveen]

Hi,

I've got (at least) one photo taken while the camera was moving.
I'd really like to get rid of the vageness caused by this motion.

I *know* it should be possible to get rid of this vagueness, but
I've never actually seen a programme that does it (only the 'sharpen'
options in the Gipm, xv, etc, but that's not what I mean).

Does anybody know of a programme (within or outside debian, but
please let it be free) that can help me?

Thanks!

joost witteveen, [EMAIL PROTECTED]

Re: hda1 problem

[Joost Witteveen]

> Hello:
> 
> I had a nicely working system with
> hda1  1 gig /
> hda2  48M   swap
> hda3  1 gig formatted space as /extra
> hda4  extended
> hda52 gigs formatted space unmounted
> 
> The root partition was getting used up so
> I tried to put /usr on hda3 using  
> 'cp -dpRv * /extra' and it appears it went 
> successfully.  

Note that that command also copyies everything that used
to be in /extra to /extra/extra -- so your new /extra may
be fuller than your original / (hda1).

> I changed the original /usr
> to /usr1 and then /extra to /usr.  Now I have
> two problems:
> 
> 1. xdm won't let me log in as anything but root
> 2. hda1 partition now shows 0 MB available and 
>100 percent usage.  du shows the same amount
>of space being used in /usr and the copied 
>partition.

Every ext2 partition has some extra space reserved for root.
So, if nobody but root can login, that usually meens the partition
really _is_ full, except for that little (5% usually, configurable)
bit of space reserved for root.

 
> Changing back to the original /usr doesn't solve 
> either problem.
> 
You didn't tell us how much space was used on the original / partition.
The way I guess now, is that both partitions were just about to be
completely full anyway, and by some coincidence the moment when both
(or rather the original /, and thus also /extra, due to the copy command)
filled up coincided with the moment when you decided to start doing
something about those space-problems, i.e. all the actions you describe
above.

> Sure could use a pointer or two while I look for
> my own solution that could make the problem much
> worse.

Well, you could try to free up some space :).


--  
Unsubscribe?  mail -s unsubscribe [EMAIL PROTECTED] < /dev/null

Re: Ethernet Question

[Joost Witteveen]

> Hi Everyone, 
> 
> I've managed to get tcp/ip networking up and running on my Debian machine. 
> I have my 3com 3x509 configured to ip address 192.168.0.1, I have one other
> machine on my local network(for now) and it is ip 192.168.0.2. 
> 
> How can I get my debian machine to see(ping) the other machine? 
> 
> My hub shows activity but ping just sits there like it is receiving nothing. 
> 
> Does this have something to do with the routing table? I've been studying a
> little and the more I read about the routing tables the more confused I get. 

You could help us help you by showing what your routing tables currently
look like. But on the other hand, I can probably help you by showing
what my routing tables look like:

$ /sbin/route -n
Kernel IP routing table
Destination Gateway Genmask Flags Metric RefUse Iface
192.168.1.0 0.0.0.0 255.255.255.0   U 0  01 eth0
127.0.0.0   0.0.0.0 255.0.0.0   U 0  01 lo

$ /sbin/ifconfig eth0
eth0  Link encap:Ethernet  HWaddr 48:54:E8:2A:77:73  
  inet addr:192.168.1.2  Bcast:192.168.1.255  Mask:255.255.255.0
  UP BROADCAST RUNNING MULTICAST  MTU:1500  Metric:1
  RX packets:1532 errors:0 dropped:0 overruns:0 frame:0
  TX packets:1644 errors:0 dropped:0 overruns:0 carrier:0
 collisions:0 
  Interrupt:9 Base address:0xdf00 

$ cat /etc/init.d/network 
#! /bin/sh
ifconfig lo 127.0.0.1
route add -net 127.0.0.0
IPADDR=192.168.1.2
NETMASK=255.255.255.0
NETWORK=192.168.1.0
BROADCAST=192.168.1.255
GATEWAY=
ifconfig eth0 ${IPADDR} netmask ${NETMASK} broadcast ${BROADCAST}
route add -net ${NETWORK}
[ "${GATEWAY}" ] && route add default gw ${GATEWAY} metric 1


If this doesn't help you, show us the output of those commands,
and we'll help you!


[EMAIL PROTECTED] / [EMAIL PROTECTED]





--  
Unsubscribe?  mail -s unsubscribe [EMAIL PROTECTED] < /dev/null