Trying to monkey-patch C is scary/awesome.
…ock now References #11.
See "Requiring other files relative to each other" in http://weblog.rubyonrails.org/2009/9/1/gem-packaging-best-practices for more info.
…be in lib/ For more info, see http://weblog.rubyonrails.org/2009/9/1/gem-packaging-best-practices ("Depending on files outside the load path" section), as well as http://github.com/defunkt/rip/issues/#issue/25
…registered regexps, instead of trying every possible combination. Closes #5. The previous version made matching regexps easier, but the number of permutations required grows too quickly: it's O(n!) for n params, which is so many that FakeWeb would seem to hang for n > 6 or so.
…y params References gh-4.
RightHttpConnection monkeypatches lots of Net::HTTP without delegating to other libraries, therefore replacing a few of FakeWeb's monkeypatches if you load it afterwards. This actually breaks Net::HTTP completely. We can probably work around this by faking at the socket level, instead of replacing Net::HTTP's #request. In the meantime, hopefully this will prevent some headaches for users of both libraries. Thanks to Ben Brinckerhoff for the report.
FakeWeb stops Samuel (and other "around advice" Net::HTTP patchers) from working when required after them, since the whole point of FakeWeb is to stop faked requests from propagating through to Net::HTTP. Everything works fine when you require FakeWeb first, though, since in that case Samuel is wrapping FakeWeb instead of the other way around.