Re: [Pyunit-interest] Unitesting organisation

[Joel Q. <joe...@ho...>]

You are right for the testing method.

I think I need to write an other configParser because, url are specify in
it.  These url are used by some other script download web page.
I think to use an other configParser with url pointing to a local web site
on my machine to provide test page.

What do you think about that phylosophy ?

Joel
----- Original Message -----
From: "Ype Kingma" <yk...@xs...>
To: "Joel Quinet" <joe...@ho...>
Sent: Saturday, October 27, 2001 10:42 AM
Subject: Re: [Pyunit-interest] Unitesting organisation


Joel,

>Hi all,
>
>I am new to unitTesting practice, I start using it after having red some
>article on
>Extreme Programming.  I have start to implement it in an application I am
>currently
>writing.  I need to do that to be able to improve it quicker, avoid
spending
>time in
>debugging due to change.  I ask myself many question about that.
>
>I have an application in Python to download web pages and extract some data
>from them.
>
>At starting point, I have used a configfile with the configParser Python
>object.
>Due to many value are not changing frequently (principaly url), I have put
>those
>variable directly in the Python code.  Now, I have the following solution,
a
>configfile
>and an object (which use the configParser object to parse the file)
>containing  more
>constant variable in it.  I have also implemented the singleton pattern in
>this object,
>this permit to avoid the parsing of the config file every time.
>
>I have implemented the folowing stuff:
>- I have put my test file in the same folder as the code file.
>- I use the httpServer in python to have a web server for testing
>
>My interrogation are :
>- should I implement a testing configfile and a testing parser object which
>contains variable too ?

That depends on what you want to test.
In case you want to test that your code is using the right
configuration you will need a separate testing configuration to compare to.

However, there is no point in writing another configuration parser.
You can test your configuration parser by feeding it a known configuration
and
compare it's output to some expected parsed configuration data.

> - What is the best solution to avoid name file conflict ?

Use a naming convention for your files. Eg. for testing cfg.py you can use
cfgtests.py, that is systematically add a 'tests' suffix to the name.
You can also put the test files in a separate directory.
Such measures will allow you to collect many unittests lateron easily
into a longer regression test.

>- Should I implement other way to put my testfile ?

I'm not sure what you mean here, but I may have answered this question
already.

>This is my current question.  I hope to be clear enough (English is not my
>mother language).

Pas de problème.

Have fun,
Ype