On 8/14/13 1:40 PM, "Justin Mclean" <jus...@classsoftware.com> wrote:
>Hi, > >Still looking into the DG issue. >1. The tests are probably wrong or at least unconventional. Tests are >passing an array of numbers as the dataProvider of an editable DG with a >label function. I agree that it might be unconventional, but it could be a valid scenario and the test failures are warning that some customer might also get busted by these changes. >2. Looks like there are several place where the code assumes data is an >object and not a String, Number (or any other type) there a few if checks >that fail as it not >checking for null but doing a straight "if (data)". I seem to recall a >few other JIRA issues around this. For sure, which is why we never tried to fix the non-complex object scenario. You are welcome to keep trying, but it might be best at this point to revert and work locally. And one thought I just had was to try to make an IList wrapper that would convert the scalar values into an object of one property. IMO, this is a corner case: most folks have complex data objects, and the few to want to use scalar values might just have to wrap their data. >3. Mustella DG test take ages to run. Yes they do, so I generally run a subfolder of tests. Then run the full set once the tests I'm worried about pass. You can also use the -caseName option to run a specific test and use -caseName more than once to run more than one test. And in this case, if you save away the failures.txt, you can restore it and run with the -failures option to keep running just those failures until they pass, then run the full DG run "later". -Alex
