On Wed, 01 Feb 2012 00:47:30 -0800, Jameson Graef Rollins <jrollins@finestructure.net> wrote: > On Wed, 1 Feb 2012 11:19:54 +0400, Dmitry Kurochkin <dmitry.kurochkin@gmail.com> wrote: > > The down side of this approach is that diff argument order depends on > > test_expect_equal_file() argument order. So sometimes we get diff > > from expected to actual results, and sometimes the other way around. > > But the files are always named correctly. > > Actually, I think this last point is the most important thing to retain. > Consistency in the diffs makes reading test results much more efficient. > The order I don't much care about. But seeing as we have been > consistent with a particular order for a while, it seems like more > effort than it's worth to change it. > It is not true that we are consistent with test_expect_equal_file() argument order. If we were, I would not bother. The problem is we are not. I remember that we already fixed argument order for test_expect_equal() and/or test_expect_equal_file(). If we do not solve this problem, we should make it a tradition. Consistent diff would be good. But IMO the current situation is worse: we are *supposed* to have consistent diff output, but in reality we have messed diff output. Also please consider the following points: * Usually one is looking at a single failing test. So it is not like you have a series of inconsistent diffs. * I personally can not remember the argument and diff order. So each time I need to understand the diff, I look at the beginning to see which side is where anyway. So IMHO diff order is not that important. But I would like to see a better solution. Perhaps Tomi's proposal would be the one. Regards, Dmitry > jamie.