[TIP] A rare philosophical thought

Carl Trachte ctrachte at gmail.com
Fri Aug 1 20:58:09 PDT 2008


On 8/1/08, C. Titus Brown <ctb at msu.edu> wrote:
> On Sat, Aug 02, 2008 at 12:59:29PM +1000, Ben Finney wrote:
>  -> "Noah Gift" <noah.gift at gmail.com> writes:
>  -> >  What is wrong with a plain assert?
>
> ->
>  -> Nothing, as far as it goes. The main improvements people usually seem
>  -> to want over a plain assert is: standard rich failure messages for
>  -> common assertions, saying what was expected; and the addition of
>  -> fixtures and other improvements to approach the Don't Repeat Yourself
>  -> ideal.
>
>
> Aaaaaaand... the real killer is that asserts are disabled in optimized
>  code, so if you use assert and want to run your tests under
>  optimization, they will not test anything.
>
>  This is actually the one thing about the current assert statement for
>  which I don't see a simple workaround.  Either you have to introduce a
>  new keyword (hah!) or a new command-line option, which I think is also a
>  bad idea.
>
>  And let me say,
>
>  I HATE UNITTEST.  UNITTEST IS EVIL.  UNITTEST NEEDS TO BE REPLACED.
>
Hey!  I'm just getting started with unittest - can we at least wait
until I learn it and nose starts working with IronPython before it
goes away . . .?

I don't like the way doctest looks - this is so superficial - but
still, Python is supposed to be all about beauty and readability.  I
agree that it is a good tool, but can't help feeling it's ugly - the
decimal module had a bunch of code that doctest can use - it just
seems to be too much.

>  There.  I feel better now.
>
>  cheers,
>  --titus
>  -----------------------------
>
>  Dr. C. Titus Brown, ctb at msu.edu
>  Asst Prof., Michigan State U.
>  http://ged.msu.edu/
>
>
>  _______________________________________________
>  testing-in-python mailing list
>  testing-in-python at lists.idyll.org
>  http://lists.idyll.org/listinfo/testing-in-python
>

Maybe it's just my (limited) experience - I have to agree most with
Jesse above - user behavior is insidiously unpredictable.  Test the
obvious cases, test the corner cases, test, test test . . .

My 2 cents . . .



More information about the testing-in-python mailing list