12 Jul
2003
12 Jul
'03
10:56 p.m.
Hi,
I hesitate to ask this here, since it's gotta be a basic shell thing, and
certainly not audio specific, so if you want to flame me for me OT and
laziness, go ahead.
The problem is that when I run I run an X application from a console (for
example, "pd &" or "rosegarden &", the console output from all
the
applications I run are interleaved onto the xterm that I started them from.
That part is fine.
The part that is not fine, is that the output is not labeled with the
application that it came from.
For example, I seem to remember that if in the past, if you ran grep, for
example, anything that grep sent to its console would be prefixed by "grep:".
I thought that this was a convention that the apps followed, and not a shell
thing (I seem to remember "printf("%s: ...", argv[0], ....")
Hopefully, I'm wrong about this, and there is a shell option that puts the
application name in front of every line of output. Is there?
If it's not something I can make happen in the shell, then how is everyone
dealing with this? Currently I am actually opening a seperate xterm for each
X application that I run, which is quite cumbersome, because it means that I
have to manage twice minus one as many windows as I would if I could have all
the output into one xterm, identified with where it came from.
Looking forward to any ideas and any pointers to my misconceptions.
Larry Troxler
12 Jul
12 Jul
11:56 p.m.
Larry:
Each process that is running has 3 devices:
Device 0: standard input: the keyboard
Device 1: standard output: the monitor
Device 2: standard error output : somehow goes to the monitor.
To trap error messages you need to redirect the standard error output
somewhere, for example:
printtool 2>/tmp/prt.log &
netscape 2>/tmp/net.log &
That is:
[command and options here] 2> log_filename_here
The ampersand will send the command to the background.
You can create a shell script with several lines and run it from a link
to it on the desktop.
Then you can check the log files with cat, more, less, etc, etc.
If you want to run several commands from one xterm session, you can do
the following:
printtool 2>&1 | sed 's/^/printtool: /g' &
netscape 2>&1 | sed 's/^/netscape: /g' &
gimp 2>&1 | sed 's/^/gimp: /g' &
2>&1 means to redirect the standard error output to the standard output
( Device 2 to device 1).
Each error line will be preceded by the command name.
I hope this will help you.
GZS
Larry Troxler wrote:
Hi,
I hesitate to ask this here, since it's gotta be a basic shell thing, and
certainly not audio specific, so if you want to flame me for me OT and
laziness, go ahead.
The problem is that when I run I run an X application from a console (for
example, "pd &" or "rosegarden &", the console output from
all the
applications I run are interleaved onto the xterm that I started them from.
That part is fine.
The part that is not fine, is that the output is not labeled with the
application that it came from.
For example, I seem to remember that if in the past, if you ran grep, for
example, anything that grep sent to its console would be prefixed by "grep:".
I thought that this was a convention that the apps followed, and not a shell
thing (I seem to remember "printf("%s: ...", argv[0], ....")
Hopefully, I'm wrong about this, and there is a shell option that puts the
application name in front of every line of output. Is there?
If it's not something I can make happen in the shell, then how is everyone
dealing with this? Currently I am actually opening a seperate xterm for each
X application that I run, which is quite cumbersome, because it means that I
have to manage twice minus one as many windows as I would if I could have all
the output into one xterm, identified with where it came from.
Looking forward to any ideas and any pointers to my misconceptions.
Larry Troxler
--
This Message is sent from my Red Hat 8.0 Linux server
.-.
.-. | Gustavo Zamorano S.
13 Jul
13 Jul
12:40 a.m.
On Saturday 12 July 2003 21:43, Gustavo Zamorano S. wrote:
Larry:
Each process that is running has 3 devices:
Device 0: standard input: the keyboard
Device 1: standard output: the monitor
Device 2: standard error output : somehow goes to the monitor.
To trap error messages you need to redirect the standard error output
somewhere, for example:
printtool 2>/tmp/prt.log &
netscape 2>/tmp/net.log &
Well, it's not just the standard error, but the standard (console) output that
I'm interested in.
You can create a shell script with several lines and
run it from a link
to it on the desktop.
Then you can check the log files with cat, more, less, etc, etc.
Yeah, but I I don't need log files - it's enough for me to just see the output
of my programs in an xterm.
If you want to run several commands from one xterm
session, you can do
the following:
printtool 2>&1 | sed 's/^/printtool: /g' &
netscape 2>&1 | sed 's/^/netscape: /g' &
gimp 2>&1 | sed 's/^/gimp: /g' &
2>&1 means to redirect the standard error output to the standard output
( Device 2 to device 1).
Each error line will be preceded by the command name.
Well, yes, this would work, but it seems very brain-damaged. I mean, do people
really use this approach to keep their output seperate?
I find it hard to believe that this is the answer. I can't imagine that it's
common practice for everyone to pipe into sed every time they run an X
program.
Larry
2:03 p.m.
On Sat, Jul 12, 2003 at 10:26:45 -0400, Larry Troxler wrote:
I generally keep each app on a differnt xterm.
gimp
2>&1 | sed 's/^/gimp: /g' &
2>&1 means to redirect the standard error output to the standard output
( Device 2 to device 1).
Each error line will be preceded by the command name.
Well, yes, this would work, but it seems very brain-damaged. I mean, do people
really use this approach to keep their output seperate? I find it hard to believe that this is the answer. I
can't imagine that it's
common practice for everyone to pipe into sed every time they run an X
program.
Well would could create start-gimp etc. scripts and put them in
/usr/local/bin. I start jack that way because it has so many arguments.
- Steve
7:43 a.m.
On Sat, 12 Jul 2003 20:43:16 -0500, Gustavo Zamorano S. wrote / a écrit:
printtool 2>&1 | sed 's/^/printtool:
/g' &
netscape 2>&1 | sed 's/^/netscape: /g' &
gimp 2>&1 | sed 's/^/gimp: /g' &
2>&1 means to redirect the standard error output to the standard output
( Device 2 to device 1).
very interesting for me, thanks.
GuyCLO~
-----------------------------------------------
#219055 http://counter.li.org
-----------------------------------------------
http://perso.wanadoo.fr/guy.clotilde/index.html
-----------------------------------------------
3:31 p.m.
Yes. it is a little OT but everyone is welcome. After all, this group
is on Linux, Audio and users.
GZS
Guy Clotilde wrote:
On Sat, 12 Jul 2003 20:43:16 -0500, Gustavo Zamorano
S. wrote / a écrit:
--
This Message is sent from my Red Hat 8.0 Linux server
.-.
.-. | Gustavo Zamorano S.
printtool 2>&1 | sed 's/^/printtool:
/g' &
netscape 2>&1 | sed 's/^/netscape: /g' &
gimp 2>&1 | sed 's/^/gimp: /g' &
2>&1 means to redirect the standard error output to the standard output
( Device 2 to device 1).
very interesting for me, thanks.
GuyCLO~
-----------------------------------------------
#219055 http://counter.li.org
-----------------------------------------------
http://perso.wanadoo.fr/guy.clotilde/index.html
-----------------------------------------------
15 Jul
15 Jul
3:21 a.m.
New subject: [linux-audio-user] Trying to compile envy24control?
I downloaded all files for ALSA release 0.9.5. They are:
alsa-driver-0.9.5
alsa-lib-0.9.5
alsa-tools-0.9.5
alsa-utils-0.9.5
You probably downloaded the tar file that is under the directory
"other". I found that it is obsolete. Running configure will fail with
alsound libs not found.
What I did is to look for the envy24control directory under
alsa-tools-0.9.5. Then, on that directory I run:
./configure
make
It will compile it successfully.
I was not able to try it because I have another sound card.
Remember to run make install.
Give it a try.
--
This Message is sent from my Red Hat 8.0 Linux server
Machine Registration number 205692 at http://counter.li.org/
.-. Gustavo Zamorano S.
.-. | Registered Linux user number 320898
.-. | | at http://counter.li.org/
13 Jul
13 Jul
12:26 a.m.
The part that is not fine, is that the output is not
labeled with the
application that it came from.
For example, I seem to remember that if in the past, if you ran grep, for
example, anything that grep sent to its console would be prefixed by "grep:"
Ewww... that would be terrible if I wanted to capture the output of a program
for later processing :)
No, you're thinking of stderr output, which by convention is
often prefixed with '$0:'. I mean, how would you like it if
you typed 'ls', only to get back:
ls: slow.wav
ls: slow.wav2
ls: smc_router_firmware
ls: sober
ls: sound
ls: soundon.log
ls: spell
ls: spell.bak
ls: src
etc... No, that's too Microshaft. Here is a possible solution:
Create an executable shell script in your path that looks like this:
#***** SCRIPT *****
function showme(){
while read foo
do echo $1: $foo
done
}
$@ | showme $1
#***** /SCRIPT *****
I called mine, "identify".
Now, let's test it with grep:
sinewave:toby:toby> identify grep foo *
grep: Binary file DSC00326.TIF matches
grep: Binary file QtC-0.0.2.lsm matches
grep: adf: afsdadsfadfasdfasdfasdfasdfasdfasdf toby adfadfadf foo
grep: adf: print('myfoot', 'myfoot');
grep: adf: print('yourfoot', 'yourfoot');
grep: doit: while read foo
grep: doit: do echo $1: $foo
So, now stdout and stderr get prefixed with the name
of the program. Easy, yes?
Tobiah
12:50 a.m.
On Saturday 12 July 2003 22:13, toby(a)tobiah.org wrote:
< I mean, how would you like it if
you typed 'ls', only to get back:
ls: slow.wav
...
Ahh, I see your point now.
Create an executable shell script in your path that looks like this:
#***** SCRIPT *****
function showme(){
while read foo
do echo $1: $foo
done
}
$@ | showme $1
#***** /SCRIPT *****
I called mine, "identify".
Now, let's test it with grep:
sinewave:toby:toby> identify grep foo *
grep: Binary file DSC00326.TIF matches
grep: Binary file QtC-0.0.2.lsm matches
grep: adf: afsdadsfadfasdfasdfasdfasdfasdfasdf toby adfadfadf foo
grep: adf: print('myfoot', 'myfoot');
grep: adf: print('yourfoot', 'yourfoot');
grep: doit: while read foo
grep: doit: do echo $1: $foo
So, now stdout and stderr get prefixed with the name
of the program. Easy, yes?
Tobiah
Well, ok, maybe easy, but still ...
I guess the answer to my original question is that, yes, people fire up a
seperate xterm for each audio app they launch :-)
I mean, I have a hard time believing that what you're describing is common
practice.
Thanks for your time! (really, thanks!)
Regards
Larry Troxler
1:58 a.m.
OH! You also want the std. output.
I read the shell given by toby(a)tobiah.org and it is also a good solution.
You can pipe things to sed, awk which are the most commonly used
commands for editing the std. output using pipes. That is, without
using temp files.
As you can see, Toby's example does not use sed, awk. It uses echo. But,
he has to use a while loop.
So, you can have:
A 4 line shell script called runit :
# runit command options
cmd=$1
# In case $1 has sub directories, then get just the command name
cmd=`basename $cmd`
#
expr="s?^?${cmd}: ?"
eval `echo "$* 2>&1 | sed '$expr' " `
This script can be installed under /usr/bin .
To run a command type:
runit [command] [command_options]
Standard output and std. errors will be displayed on the xterm with
command name included.
After all, what I love in Unix is that YOU CAN do these kind of things.
Everything is possible.
I recognize that to be able to do this 4 line script takes too much Unix
experience.
I am using basename to get the filename in case you run:
runit /usr/local/bin/netscape &
I define expr to create the argument sed expects. I use quetion marks
to avoid conflicts with slashes.
Life is harder on any other MOST popular operating systems...specially
those that crash so often in a daily basis.
GZS
Larry Troxler wrote:
On Saturday 12 July 2003 22:13, toby(a)tobiah.org
wrote:
< I mean, how would you like it if
--
This Message is sent from my Red Hat 8.0 Linux server
.-.
.-. | Gustavo Zamorano S.
you typed 'ls', only to get back:
ls: slow.wav
...
Ahh, I see your point now.
Create an executable shell script in your path
that looks like this:
#***** SCRIPT *****
function showme(){
while read foo
do echo $1: $foo
done
}
$@ | showme $1
#***** /SCRIPT *****
I called mine, "identify".
Now, let's test it with grep:
sinewave:toby:toby> identify grep foo *
grep: Binary file DSC00326.TIF matches
grep: Binary file QtC-0.0.2.lsm matches
grep: adf: afsdadsfadfasdfasdfasdfasdfasdfasdf toby adfadfadf foo
grep: adf: print('myfoot', 'myfoot');
grep: adf: print('yourfoot', 'yourfoot');
grep: doit: while read foo
grep: doit: do echo $1: $foo
So, now stdout and stderr get prefixed with the name
of the program. Easy, yes?
Tobiah
Well, ok, maybe easy, but still ...
I guess the answer to my original question is that, yes, people fire up a
seperate xterm for each audio app they launch :-)
I mean, I have a hard time believing that what you're describing is common
practice.
Thanks for your time! (really, thanks!)
Regards
Larry Troxler
14 Jul
14 Jul
1:26 p.m.
Am Sonntag, 13.07.03, um 02:43 Uhr (Europe/Berlin) schrieb Larry
Troxler:
Hi,
I hesitate to ask this here, since it's gotta be a basic shell thing,
and
certainly not audio specific, so if you want to flame me for me OT and
laziness, go ahead.
The problem is that when I run I run an X application from a console
(for
example, "pd &" or "rosegarden &", the console output from
all the
applications I run are interleaved onto the xterm that I started them
from.
That part is fine.
The part that is not fine, is that the output is not labeled with the
application that it came from.
For example, I seem to remember that if in the past, if you ran grep,
for
example, anything that grep sent to its console would be prefixed by
"grep:".
I thought that this was a convention that the apps followed, and not a
shell
thing (I seem to remember "printf("%s: ...", argv[0], ....")
Hopefully, I'm wrong about this, and there is a shell option that puts
the
application name in front of every line of output. Is there?
Unfortunately not. The programs just write to stdout/stderr and the
shell, being just another program
does not even get to see this output. So the only chance is, that the
programs do it themselves.
Otoh doing just printf("%s: ...", argv[0], ....") is probably no what
most users want.
Imagine the output being saved to a file or piped to another program.
So if a program wants to offer this service it should check, if its
writing to a tty
(unix speak for any kind of terminal) and most of the programmers did
not
make this effort for every printf (in pseudo C)
if(isatty(stdout))
myname=argv[0];
else
myname="";
printf("%s: ...", myname, ....")
If it's not something I can make happen in the
shell, then how is
everyone
dealing with this? Currently I am actually opening a seperate xterm
for each
X application that I run, which is quite cumbersome, because it means
that I
have to manage twice minus one as many windows as I would if I could
have all
the output into one xterm, identified with where it came from.
If you need the output, save it to a file. Dont forget stderr:
pd >& ~/pd.out &
(csh-like)
regards,
Thomas N.
7739
days inactive
7742
days old
linux-audio-user@lists.linuxaudio.org
11 comments
6 participants
participants (6)
-
Gustavo Zamorano S. -
Guy Clotilde -
Larry Troxler -
Steve Harris -
Thomas Neuhaus -
toby＠tobiah.org