Re: isnanl: recognize non-IEEE numbers

[Bruno Haible]

James Youngman wrote:
> It might be more helpful to use meaningful variable names for your
> constants here, because that way the error message you get when the
> assert fails tells you which test has gone wrong.

But in order to understand the test, you have to look at the test's source
code anyway.

> Otherwise, you will
> have to compare line numbers, and risk wasting time due to line number
> skew between yourself and a possible problem-reporter.

This risk is low, because the reporter has to specify the package he is
testing anyway: be it findutils-4.3.6, or m4-1.4.9b, or the most recent
gnulib. We need this information from every reporter anyway, otherwise
we start hunting bugs that are already fixed, and so on.

Bruno