Friday, December 22, 2006

Reload modules in Python environment

You create a module and run it using the python interpreter. Now you modify this module and try to run it again, surprise surprise, the new changes are not reflected.

What really happens:
The python interpreter loads modules as and when they are used in a script. Once these modules are loaded they are cached in memory for reuse later. Now if you modify a pre-loaded module and try to run it, the interpreter does not bother loading it again as it already has a cached copy.

One technique that i have used effectively for (indirectly) reloading the modules is to delete the module object from the cache. This forces the interpreter to reload the module when it executes the script again and you can test your new changes easily.

How its done:
The interpreter caches the module objects in a PyDictionary (a map) called the sys.modules. This map stores the module objects in the following format:
key : the fully qualified class/module name
value : the actual module object

The basic trick here is to delete the required object from this map and your jobs done.


if "<the fully qualified class/module name>" in sys.modules.keys():
<the fully qualified class/module name>"]

line 1 : test if the module exists in the cache
line 2 : actually remove the module from the map based on its key

Say, you want remove the module "com.test.testfile" from the cache.

if "com.test.testfile" in sys.modules.keys():

Put these lines of code after the import statements or before you make a reference to your changed module, this ensures that the module is reloaded before execution.

There may be better ways of reloading modules but i found this most effective when working with a system which contains a lot of scripts.

Tuesday, December 05, 2006

bug tracking tool

right so the last 2 days i have been working on setting up a bug tracking tool for my boss. Some how i knew that one of these days the idea of acting smart was going to backfire. ;) ;) So here was my boss telling me how he was confident that i would help him out with setting up a bug tracking tool for his new project.

nways i got down to the task and did a bit of R&D on the different issue/bug tracking tools and here's what i found:

jira - if you have the cash, the best tool in the market

bugzilla - the king of open source and free tools, its heavily loaded but complex to setup, plus its written in perl

roundup - python based and quite good

But the ones i picked :

mantis - PHP based but simple to work with, its also called the unofficial 'light-weight' bugzilla

JTrac - Java based, not very impressive UI but ideal if your target audience is small. Best part is that its demo version is quite good. It has embedded jetty web-server and HSQLDB at the backend. You can be up and running in minutes, plus maintaining it is quite simple.

i'm sure you guessed what i used to impress my boss. Simple but working is still the flavour of the day. :)