From: Rick M. <obj...@gm...> - 2008-09-04 16:23:48
|
I just committed a cleaned up version of the abs bif tests. Running these in both 9 and 18 really seems the correct thing to do. Rick On Thu, Sep 4, 2008 at 11:37 AM, Mark Miesfeld <mie...@gm...> wrote: > On Thu, Sep 4, 2008 at 8:23 AM, Rick McGuire <obj...@gm...> wrote: >> Some initial impressions on scanning the logs. >> >> Most of the tests have "expected" and "actual" reversed on the asserts :-) > > Yes. ;-) The failures are certainly easier to read if expected and > actual are in the right order. There was a disconnect in this when > the framework was first being developed and it was hard to remember > which arg was which. > > Now the rule to follow is: Where there is an "expected" and "actual" > arg, "expected is always the first arg. > >> It's pretty easy to pick out what's going wrong from the logs, and >> generally things don't look too bad. In generaly, the test groups >> with the largest numbers of failures tend to be the arithmethic >> operations, and that's because the same operation under numeric digits >> 9 and numeric digits 18 would generally produce different results >> because there's more significant digits. I think for these tests, I'd >> prefer that these test variables have both a digits 9 and a digits 18 >> variation and both get run on all platforms. Since the precision is a >> controllable item here, I think we'd have better validation if >> everything is run both ways on all platforms. > > Okay, that makes sense. > >> Of what's left, I don't see there's any need for creating new test >> groups. There tend to be just a few cases scattered around that can >> be easily dual pathed...or in some cases, the test case data can be >> changed to be independent. A good example of this last category is a >> failure in the copies test group. It's using the digits() bif as the >> source of the copied data, and doesn't handle the change from 9 to 18 >> particularly well. There are probably better ways of doing that >> particular test. > > Good. Fixing the test cases in this last category is useful. Test > cases should be independent of things that can vary from system to > system, if at all possible, to begin with. > > -- > Mark Miesfeld > > ------------------------------------------------------------------------- > This SF.Net email is sponsored by the Moblin Your Move Developer's challenge > Build the coolest Linux based applications with Moblin SDK & win great prizes > Grand prize is a trip for two to an Open Source event anywhere in the world > http://moblin-contest.org/redirect.php?banner_id=100&url=/ > _______________________________________________ > Oorexx-devel mailing list > Oor...@li... > https://lists.sourceforge.net/lists/listinfo/oorexx-devel > |