pty.spawn directs stderr to stdout

[Wilbert Berendsen]

Hi,

using pty.spawn() it seems that stderr output of the spawned process is
directed to stdout. Is there a way to keep stderr separate and only direct
stdin and stdout to the pty?

TIA,
w best regards,
Wilbert Berendsen

--
http://www.wilbertberendsen.nl/
"You must be the change you wish to see in the world."
-- Mahatma Gandhi

[Donn Cave]

In article <ma************ *************** ***********@pyt hon.org>,
Wilbert Berendsen <wb****@xs4all. nlwrote:

Hi,

using pty.spawn() it seems that stderr output of the spawned process is
directed to stdout. Is there a way to keep stderr separate and only direct
stdin and stdout to the pty?

There is, of course.

First, you have to decide where you want unit 2 ("stderr") to go, and
then get the spawned process to redirect it there. If a disk file
will do, then your question is just "how do I redirect error output
to a disk file, in ___" (fill in the blank with language used to
implement the spawned process - UNIX shell? Python? C?)

More likely, you want the spawned process' error output to go wherever
the parent's error output was going. This is a little trickier.

Ideally, your spawned shell script can conveniently take a new
parameter that identifies the new file descriptor unit number for
error output. In this case, use fd2 = os.dup(2) to get a new
duplicate, add a parameter like -e str(fd2), and in the spawned
process, redirect from that unit - in UNIX shell, exec 2>&$fd2

Or you could use an environment variable to identify the backup
error unit, if the command line parameter option isn't available
for some reason.

Donn Cave, do**@u.washingt on.edu

[Donn Cave]

In article <ma************ *************** ***********@pyt hon.org>,
Wilbert Berendsen <wb****@xs4all. nlwrote:

Hi,

using pty.spawn() it seems that stderr output of the spawned process is
directed to stdout. Is there a way to keep stderr separate and only direct
stdin and stdout to the pty?

There is, of course.

First, you have to decide where you want unit 2 ("stderr") to go, and
then get the spawned process to redirect it there. If a disk file
will do, then your question is just "how do I redirect error output
to a disk file, in ___" (fill in the blank with language used to
implement the spawned process - UNIX shell? Python? C?)

More likely, you want the spawned process' error output to go wherever
the parent's error output was going. This is a little trickier.

Ideally, your spawned shell script can conveniently take a new
parameter that identifies the new file descriptor unit number for
error output. In this case, use fd2 = os.dup(2) to get a new
duplicate, add a parameter like -e str(fd2), and in the spawned
process, redirect from that unit - in UNIX shell, exec 2>&$fd2

Or you could use an environment variable to identify the backup
error unit, if the command line parameter option isn't available
for some reason.

Donn Cave, do**@u.washingt on.edu

[Wilbert Berendsen]

Op vrijdag 11 april 2008, schreef Donn Cave:

More likely, you want the spawned process' error output to go wherever
the parent's error output was going. *This is a little trickier.

I ended up writing a small script that basically reimplements fork() from the
pty module, where then STDERR is not dup'ed to the pty.

I'm currently trying to combine subprocess.Pope n with pty.openpty ...

tx,
Wilbert Berendsen

--
http://www.wilbertberendsen.nl/
"You must be the change you wish to see in the world."
-- Mahatma Gandhi

[Wilbert Berendsen]

Op zaterdag 12 april 2008, schreef Wilbert Berendsen:

I'm currently trying to combine subprocess.Pope n with pty.openpty ...

which succeeded :-)
http://code.google.com/p/lilykde/sou...unpty.py?r=314
it seems to work :-)
thanks,
Wilbert Berendsen

--
http://www.wilbertberendsen.nl/
"You must be the change you wish to see in the world."
-- Mahatma Gandhi