Am 10.12.2009 um 17:22 schrieb Paul Vriens: > Well, the patch has been committed but the ok() message looks a bit strange > now. Do you think it makes sense to change getdc_capable to a HRESULT and do > something like the following instead: > > ok(hr == testdata[i].result || > testdata[i].alt_result && hr == testdata[i].alt_result, > "GetDC on a %s surface returned 0x%08\n", > testdata[i].name, hr); That code makes it a bit more obvious than the last one, but those tests are of pretty little use now. The test essentially says "it can work or it cannot work". We also have a few other tests testing and documenting the return values in specific situations(though I think not in the situation where the format cannot be used with GetDC.
I have no problems if we leave the tests as they are right now. If you want, you could disable the tests entirely, revert the alternative ok() value, and add a separate test that tests what the return value is if a format can't be used with GetDC(but use a format that fails on all windows versions, like DCT1)
