Increase your debugging and development productivity by logging copious amounts of data!
A mixed approach between debugging and logging, this package dumps your objects into an HTML+JS based browser where you can later inspect them and find problems. You can initiate a dump by inserting a call into your code or any time an exception is thrown. You can dump specific values or the entire stack frame, including up in the call stack.
Of course this can be time or memory consuming. That's why it's not recommended to be used in production at all, dumps are limited in depth and we try to filter out uninteresting files from the Python library. Even with these restrictions, I find it useful.
The HTML+JS data browser has some filtering and grouping capabilities, can filter dumps from different processes based on PID, group dumps by thread, enclosing stack frame or
The code is a bit messy, sometimes downright unmaintainable. Sorry for that, it was written in a rush :)
- Enable exception tracing and enjoy the ability to record & inspect context any time an exception is thrown.
- Something is wrong and you're too lazy to launch a debugger or type exact object
address to logging. Just drop a line of
ovlocal()in that place and look things up in the browser. - Keeping track of values between different test runs.
- Install using your favourite package manager:
pip install overlog - Run the web server:
python web/server.py - Open
http://localhost:8111/in your browser - In your code:
from overlog import ovlg, ovlocal
# now somewhere in your code:
def my_computation(rabbits, wolves):
new_rabbits -= 2*wolves
grass = 100 - 3*new_rabbits
ovlocal()
# ^ this dumps the current stack frame which includes the variables rabbits, wolves, new_rabbits and grass
# and also dumps a few stack frames up the call stack
- Switch to the browser and explore your object dump there.
The overlog library is included in your programt that needs debugging. When dumping is invoked, it creates a JSON representation of the object(s) of interest and sends them using standard HTTP to a Python Tornado server running in the background which in turn, is connected to the browser via WebSockets. The browser receives objects in JSON and displays them on the page.
Create a sitecustomize.py file and put it in your site-packages with the following
content:
try:
from overlog import ovlg
import sys
sys.ovlg = ovlg
except:
pass
and then you don't ever need to import overlog again!
I actually like to auto-run the overlog server after system boot so it's always ready to use.
This project was strongly inspired by Bret Victor, the LightTable editor and people behind it and also the https://github.com/Akson/RemoteConsolePlus3 project.
The code is LGPL because I would like to see contributions coming back but at the same time, it should not limit you because this library is not intended to be redistributed with other software.