On 2010-7-21 07:32, Wolfgang Schnerring wrote:
* Wichert Akkerman<wichert@wiggy.net> [2010-07-20 19:28]:
On 2010-7-20 18:15, Christian Theune wrote:
At least, WRT this bug, I don't think it's a good idea to ask explicitly for bad requests to go to the application as the test layer should model real server behaviour as closely as possible. And again it wouldn't make sense anyway as you can't pass an unparsable request to the application.
I'm not sure I agree. Like everything else servers have bugs, so it can't hurt to test how your application would behave given certain server bugs.
I don't think it is usually a productive assumption that lower layers fail to uphold their end of the contract. Maybe an extrapolation/hyperbole illustrates my opinion: Cosmic rays might also flip bits in your computer's RAM or disk, but I don't think it's worthwile to test how your application reacts when the python interpreter (or whoever, really) presents it with mangled data structures or objects or whatnot.
And for some situations you do want to explicitly test for such things. It all depends on how critical your app is. I'm not sure I'm not the only one who has tested code with randomly broken/bit-flipped input to test robustness. Wichert. -- Wichert Akkerman <wichert@wiggy.net> It is simple to make things. http://www.wiggy.net/ It is hard to make things simple.