Traceback in Logging

[Kottiyath]

Hi all,
Is it possible to print traceback for all logging?
I have created something like this:

def trace():
import traceback
return ''.join(traceback.format_stack(limit=4)[1:])

And during every logging, append it -
self.logger.info('Data: %s, kwargs: %s Traceback: %s' %(data,
kwargs, trace()))

-->Result<--

2009-01-06 18:52:21,483 - test- INFO
---> Data:, kwargs: {}
Traceback: File "C:\test.py", line 48, in success
super(testObj, self).success(reply, **kwargs)
File "C:\test.py", line 87, in success
self.logger.info('Data: %s, kwargs: %s Traceback: %s' %(data,
kwargs, trace()))
File "C:\test.py", line 151, in trace
return ''.join(traceback.format_stack(limit=4)[1:])

This seems somewhat kludgy. Is it possible in logging mechanism itself
to provide traceback as default?

I use Python2.4, so LoggingAdapter is not there too.

 

[Vinay Sajip]

This seems somewhat kludgy. Is it possible in logging mechanism itself
to provide traceback as default?

No, because it's not a common use case to print tracebacks for every
logging call. There's support for adding traceback information in
exception handling, via use of the Logger.exception method in the
exception handling code.

I use Python2.4, so LoggingAdapter is not there too.

If you need LoggingAdapter, you can always copy and paste the relevant
code from Python's SVN repository into your own application.

Regards,

Vinay Sajip

 

[Kottiyath]

I dont want the whole traceback. I just wanted to know where the log
was generated from i.e. which procedure and which line. I have 3/4
points in many procedures where I encounter a small error (not an
exception) and want to log it. So having individual names for each
looks to be somewhat verbose - esp since the application is >10K LOC.

This might be a good item to have in the logging system - along with
time and level, just proc name and the line number.

Thank you for the help.

Regards
K

 

[skip]

Kottiyath> I dont want the whole traceback. I just wanted to know where
Kottiyath> the log was generated from i.e. which procedure and which
Kottiyath> line.

The asyncore standard module has an undocumented compact_traceback()
function:

#!/usr/bin/env python

import asyncore

def f():
1/0

def g():
f()

def h():
g()

try:
h()
except ZeroDivisionError:
print asyncore.compact_traceback()[3]

Output looks like this:

% python f.py
[f.py|<module>|15] [f.py|h|12] [f.py|g|9] [f.py|f|6]

Might be of interest.

 

[Vinay Sajip]

I dont want the whole traceback. I just wanted to know where the log
was generated from i.e. which procedure and which line. I have 3/4
points in many procedures where I encounter a small error (not an
exception) and want to log it. So having individual names for each
looks to be somewhat verbose - esp since the application is >10K LOC.

Don't the funcName and lineno arguments in the format string work for
you?

(See http://docs.python.org/library/logging.html#id1)

Regards,

Vinay Sajip

 

[Thorsten Kampe]

* Vinay Sajip (Tue, 6 Jan 2009 11:24:54 -0800 (PST))

Don't the funcName and lineno arguments in the format string work for
you?

(See http://docs.python.org/library/logging.html#id1)

I don't think he's interested in the "line number where the logging call
was issued" but where the exception occurred.

Thorsten

 

[Vinay Sajip]

I don't think he's interested in the "line number where the logging call
was issued" but where the exception occurred.

Well, he does say "all logging" and "every logging". And, as I
mentioned earlier, there is already support for tracebacks by using
the Logger.exception method in an exception handler.

Regards,

Vinay Sajip

 

[Kottiyath]

The issue is that I am on Python 2.4 which doesnt support func name.
I am using filename and lineno now. That does serve the purpose.
Thank you, I had not checked all the parameters.

Regards
K