libgpg-error t-printf test fails on powerpc and ppc64

[NIIBE Yutaka]

Hello, again,

NIIBE Yutaka <gniibe at fsij.org> wrote:
> I pushed a fix (to libgpg-error master) before my vacation:
>
> 	9cc750a816554e46d3e1496533f927b17ef3fcef
>
> Today, I tested on perotto.debian.net and confirmed that:
>
> 	t-printf: IBM double-double - skipping LDBL_MAX test
>
> The test is skipped correctly.

I realized that my change was not good.  It only cares about 64-bit
machine (reported in July), but it is also relevant to 32-bit machine
(as you reported).

I pushed another commit to cover powerpc (32-bit):

	747eb8798bed752ede9b4cec03507e64bf6a497a

(Well, we could only test by LDBL_MANT_DIG == 106.  But I don't want to
encourage such a situation for an architecture in the future.  We will
need to change again when such a situation will occur.)
--