> Date: Wed, 29 Nov 2017 05:37:56 +0700 > From: Robert Elz <k...@munnari.oz.au> > > I think that conclusion had been reached already (not by me...) but that's > "Under IEEE 754-2008" right? What about architectures that don't use IEEE > floats? This test should not be assuming that - we still support vax right? > The test should be able to be run there as well. Can we guarantee that the > results must be identical on every port that runs NetBSD?
There's no reason the same algorithm shouldn't work on VAX too. Even if it didn't (maybe because VAX's basic arithmetic operations give spectacularly incorrectly rounded answers), I'd rather prioritize reporting bugs in what we expect to be IEEE 754 floating-point arithmetic over hiding bugs everywhere. If it turns out that this test is legitimately wrong on VAX, we can conditionalize it for VAX. > | Can you please put it back to an exact equality test and report the > | mismatched values in the output, with %a rather than (or in addition to) > %g > | so that we can see the precise difference in their bits when the test > fails? > > Yes, doing that now (need to build and test twice, once with the "correct" > code, and once where the result of the test is deliberately forced to > fail so I can test the error string works as intended.) Thanks!
