waiting for a program run by popen3() to end

[Rajarshi Guha]

Hi,
I have some code that runs an interactive program via popen3(). The program generates
output which is then used later on. My script looks like this:



o,i,e = os.popen3('/usr/local/adapt/bin/descmng -g')
o.write('\n%s\n \n' % (redpick))
o.close()
time.sleep(1)
os.system('mv set5 gendes.in && rm set? outputdesc.txt' )

The program (descmng) produces the files set?, outputdesc.txt

However on some systems, the program takes more than 1 sec to complete
and as a result the os.system() command fails.

Increasing the sleep time to 5 sec solves the problem but it does'nt seem to
be reliable.

I have considered using fork() and wait() but that does'nt let me run the
program interactively.

Is there any other way out (apart from dumping commands to a file and
piping them to my program in a fork())?

Thanks,

[Donn Cave]

Re: waiting for a program run by popen3() to end

Quoth Rajarshi Guha <rajarshi@presi dency.com>:

| I have some code that runs an interactive program via popen3(). The program generates
| output which is then used later on. My script looks like this:
|
|
|
| o,i,e = os.popen3('/usr/local/adapt/bin/descmng -g')
| o.write('\n%s\n \n' % (redpick))
| o.close()
| time.sleep(1)
| os.system('mv set5 gendes.in && rm set? outputdesc.txt' )
|
| The program (descmng) produces the files set?, outputdesc.txt
|
| However on some systems, the program takes more than 1 sec to complete
| and as a result the os.system() command fails.
|
| Increasing the sleep time to 5 sec solves the problem but it does'nt seem to
| be reliable.
|
| I have considered using fork() and wait() but that does'nt let me run the
| program interactively.
|
| Is there any other way out (apart from dumping commands to a file and
| piping them to my program in a fork())?

You should look at the popen2 module in the library. You're using
it already, via os.popen3, but the classes in popen2.py provide more
features. However, it seems to me that you should be able to synch
with the child process by simply reading from its output or error
files. Whether it actually outputs any data or not, when the process
exits, those pipes should close and the read will finish. If you're
not interested in the data, though, I would not redirect error output
and throw it away as you're doing here, and if you redirect only the
regular output, that will greatly reduce odds of a deadlock.

Or even simpler, you can add the other statements to the command
to be run by os.popen3.

Donn Cave, donn@drizzle.co m

[Miki Tebeka]

Re: waiting for a program run by popen3() to end

Hello Rajarshi,
[color=blue]
> o,i,e = os.popen3('/usr/local/adapt/bin/descmng -g')
> o.write('\n%s\n \n' % (redpick))
> o.close()
> time.sleep(1)
> os.system('mv set5 gendes.in && rm set? outputdesc.txt' )[/color]
If you're not reading the program output you can use os.system which
will wait for it to end.
[color=blue]
> Is there any other way out (apart from dumping commands to a file and
> piping them to my program in a fork())?[/color]
On Unix like systems there is popen2.Popen3 (or Popen4)
---
from popen2 import Popen3

print "Running child"
p = Popen4("my_proc ess arg1 arg2")
print "Waiting"
p.wait() # Will wait for process to end
print "And we got:"
print p.fromchild.rea d()
---

HTH.
Miki