win32pipe.popen3

Jakob Simon-Gaarde jsgaarde at gmail.com
Wed Aug 17 19:51:41 CEST 2005


Follow-up on a thread from 1999 (see below)

Well now it is 2005 and the operating system I'm using is Windows
Server 2003, and I can still see that the same problem persists with:

win32pipe.popen2()
win32pipe.popen3()
win32pipe.popen4()

while win32pipe.popen() does almost what you want.

>>> import win32pipe
>>> win32pipe.popen('cmd')
<open file 'cmd', mode 'r' at 0x009DD698>
>>> r=win32pipe.popen('cmd')
>>> r.readline()
'Microsoft Windows XP [Version 5.1.2600]\n'
>>> r.readline()
'(C) Copyright 1985-2001 Microsoft Corp.\n'
>>> r.readline()
'\n'
>>> r.readline()
'C:\\backup\\TRPython241\\trpython>'

Although I think the last readline ought to return None since no
carriage return has been issued yet, it is better than popen2,popen3
and popen4, which all just block the parent process.

The current behaviour of win32pipe.popen2(), win32pipe.popen3() and
win32pipe.popen4() would be acceptable for me if I knew a way to test
if there was something ready for reading, but I can't see how to do
that test, therfore I don't know when to stop reading from output :( Is
there a solution for this, can I poll/test for ready-read on popen3 I/O
objects.

Best regards
Jakob Simon-Gaarde


---------------------------
>From a thread in 1999
High Arpard,

thanx for help but I got probs with that popen3 under Win95:
'o.readlines()' doesn't return anymore. To find out I checked
it line per line:

Python 1.5.2 (#0, Apr 13 1999, 10:51:12) [MSC 32 bit (Intel)] on win32
Copyright 1991-1995 Stichting Mathematisch Centrum, Amsterdam
>>> import win32pipe
>>> i,o,e=win32pipe.popen3('ver', 'b')
>>> o.readline()
'\015\012'
>>> o.readline()

'Windows 95. [Version 4.00.1111]\015\012'

>>> o.readline()
'\015\012'
>>> o.readline()

Don't know why, but it never;-) returns.
Perhaps it may be a bug in win32pipe, that it doesn't return
becourse readline couldn't find CarriageReturn/EOF?

I took win32pipe.popen('ver','r') for a better solution.
That works fine.

greetings,
Holger




More information about the Python-list mailing list