From: Siddharth P. <si...@cs...> - 2008-04-10 02:11:36
|
> Argh! Just realized that SenseRelate-AllWords was using QueryData's > now deprecated version () method! So, I need to change that. We used > that in my tests in 0.08 that were attempting to get the version info > from the user, but since that isn't always reliable now, I think the > test didn't perform as expected! > > In any case, I think that's at the root of these puzzling bug reports > for 0.08 (the case of fire#n#2 - the version thing wasn't working as > expected, and I think users got the wrong expected values in some > cases...) Right. Since, WordNet::Similarity is already a dependency, you can use the hashCode method of WordNet::Tools in it. > Also, I tried changing some of the measures in the .t tests to ones > other than lesk, specifically wup, path, and lch, and there were tons > of errors. :) So, I think we want to look at that. Those actually > might be a good way to try out this "todo" type of test in > Tests::More, where you expect a test to fail so you can work on it in > future. I'll see if I can get that figured out. Hmmm... I think the failures should be expected. I think the test is basically running AllWords with the lest measure, and then comparing the output to some expected output (hard-coded in the test). So, most likely this output is different with different measures. But I think it would be good to add tests for each of the measures doing this. -- Sid. |