Re: Testsuite failures of gawk 5.2.0

[arnold]

Eli Zaretskii <eliz@gnu.org> wrote:

> Maybe I'm missing something, but why does it matter to us what is the
> sign of NaN?  If this is just for the test suite, we could accept both
> results.  If this is for some other functionality, what is that
> functionality?

Getting the test suite to accept both is a bit of a hassle. I may
change it so that this particular test doesn't fail the test suite.

It matters in the larger sense in that I'd like gawk to have consistent
behavior across platforms.  When converting the string "-nan" to a
number by adding zero to it, a "negative" NaN should result.

It's very much a quirky corner case, but it can come into play
in gawk's numeric comparison rules.

I'm still considering what to do about this.

Arnold