[Skip Montanaro] > This reminded me of something I noticed awhile ago. If you're learning > something new, there is a tendency to find a working example to start from, > then modify it to suit your needs. This is fine as far as it goes, however, > if the idioms used in the code you're cloning are suboptimal, they get > cloned. > > Unittest's API is difficult enough for me to remember (at least the initial > framework I need to put together) that I generally hunt down some previous > unittest usage, clone it and start from there. I no longer have any idea > what the original unittest example was that got me started using it in the > first place (probably something in the distro's docs). I can only hope it > was a good example.
If it was the distro or tutorial example, I hope you thought it was good. I added those examples to the docs with the intention facilitating cut-and-paste jobs so folks could get up and running quickly. The remainder of the unittest docs are somewhat obtuse and uninspiring. > >From that standpoint the simpler API of py.test seems attractive. That is their number one selling point. I hope they don't lose it as the package evolves to include doctests and such. Raymond Hettinger -- http://mail.python.org/mailman/listinfo/python-list