Question

I have a GUI python program and a simple error logging system via

import sys
sys.stderr = open("err.log", "w")

and it works mostly fine. The one problem I have is that whenever I encounter a run-time error using Windows 7 and Python 2.7.3, the file err.log gets written only after I close the program. Reading related issues I gather I’d need to perform flush() and os.fsync() to sys.stderr after the error, but I don’t know how to do that easily.

One way would perhaps be to perform flush/fsync after every possible point in the program where a run-time error is possible, but this is obviously not a good solution. The main reason for using this error logging is debugging, so by definition I can’t really know beforehand where I’d need to flush other than at all the possible places where a run-time error is possible. Since there are a lot of such places I’d prefer not to have to try/catch every one of them.

I’m using PyGTK, so there is no main loop visible to me in which I could do the flush/fsync.

Is there any way to tell Python to always perform flush and fsync after an error, or does anyone have other ideas how to solve this (minor) problem?

Answer 1

I agree with syhpoon: using logging for logging. But if you really need your arrangement, look at the third buffering argument to open to disable it:

If the buffering argument is given, 0 means unbuffered, 1 means line
buffered, and larger numbers specify the buffer size. The preferred
way to open a file is with the builtin open() function.

I tried this and it works. In a Python session:

>>> a = open('/tmp/bar', 'w')
>>> a.write('hi\n')

And in another term window:

$ cat /tmp/bar
$

Then I re-ran the experiment with buffering disabled:

>>> a = open('/tmp/bar', 'w', 0)
>>> a.write('hi\n')

And:

$ cat /tmp/var
hi
$

So use logging, but if not, see if just disabling buffering will give you what you need.