On 07/09/2010 04:58 AM, Sylvain Munaut wrote: >> (even though what could be properly decoded before >> also can be properly decoded now, no change for this). > > Well, I think that before > > encode(decode(x)) == x
Hi all, if the decoder is too much work right now, you can change the test to only test encoding. So make it encode(x) == expected_result and add one test case for the string that was failing, convert the old decoding to decode(x) == expected_result as well. and you might feel like adding an expected failure (because the code is missing). The benefit of changing the test is that you can more easily convince me that the new code can do everything the old promised and is fixing something that didn't work with the old one.