On Thu, Jun 11, 2009 at 17:01, Jeff Stoner<leapf...@freeshell.org> wrote: > On Thu, 11 Jun 2009, Gisle Aas wrote: > >> I would like to avoid this kind of test. First because that allow me >> to use the native type names directly, > > The change doesn't prevent this. You can specify either the native name or > the alias name.
I misread you line of code. I've been doing to much Python programing lately where iterating over a hash just gives you the keys :-) >> the module to support types that the web-service introduce after the >> module was released. > > How can I be a BOFH if I let people do what they want, when and how they > want. ;-) > > But I do understand this point. Making it "mostly future-proof" means less > work for users and you to maintain it. My change certainly prevents that. Changing it to just warn (with carp) instead might be an acceptable compromise. > Since it's a one line change, I can easily track the differences between my > local copy and your official distribution. > >> I don't belive passing the wrong type will be a common error. You >> will need to test the generated URLs anyway and the error will be >> discovered then. > > That's how I came up with the suggestion - I used your example in the POD. > You used "line" instead of "lines" and generated a broken api call. Oh, I'll fix that broken docs then. There was even another severe typo in that code. --Gisle