Re: isfinite: fix bugs

[Bruno Haible]

Ralf Wildenhues wrote:
> Wouldn't it be more useful if, instead of listing Linux/x86, the
> glibc version that had this bug was listed (and maybe the GCC
> version used);

It would be useful if there was some hope that glibc be more defensive
against these bit patterns that are outside the IEEE 754 concepts.

In [1], Ulrich Drepper's opinion:
  "Garbage in, whatever out, including crashes."
In [2], Jakub Jelinek's opinion is that these cases should be handled
as easily and cheaply as possible, not as defensively as possible. So if
they want even printf to show pseudo-denormals as finite numbers, because
it's cheaper, they will not want isnan or isfinite to handle these specially
either.

> also, is there a glibc bug report about this?

There was [3] but it has disappeared now.

Bruno


[1] http://lists.gnu.org/archive/html/bug-gnulib/2007-06/msg00061.html
[2] http://lists.gnu.org/archive/html/bug-gnulib/2007-06/msg00055.html
[3] http://sourceware.org/bugzilla/show_bug.cgi?id=4586