|
ecab4471
|
2009-02-26T20:18:42
|
|
avoid gcc 3.4.3 bug on long double NaN on Irix 6.5
* tests/nan.h (NaNl): Rewrite as function on Irix, to avoid
compilation bug by using runtime conversion.
* m4/isfinite.m4 (gl_ISFINITE): Likewise.
* m4/isnanl.m4 (gl_FUNC_ISNANL): Likewise.
* modules/ceill-tests (Files): Use nan.h.
* modules/floorl-tests (Files): Likewise.
* modules/frexpl-tests (Files): Likewise.
* modules/isnanl-tests (Files): Likewise.
* modules/ldexpl-tests (Files): Likewise.
* modules/roundl-tests (Files): Likewise.
* modules/truncl-tests (Files): Likewise.
* tests/test-ceill.c (main): Use a working NaN.
* tests/test-floorl.c (main): Likewise.
* tests/test-frexpl.c (main): Likewise.
* tests/test-isnan.c (test_long_double): Likewise.
* tests/test-isnanl.h (main): Likewise.
* tests/test-ldexpl.h (main): Likewise.
* tests/test-roundl.h (main): Likewise.
* tests/test-truncl.h (main): Likewise.
See http://lists.gnu.org/archive/html/bug-gnulib/2009-02/msg00190.html.
Signed-off-by: Eric Blake <ebb9@byu.net>
|
|
ea0f5bf6
|
2008-10-08T02:05:26
|
|
Use a more portable replacement expression for -0.0L.
|
|
|
5d7e7dd7
|
2007-04-06T21:15:32
|
|
Test also the sign bit of zero results.
|
|
|
7fda7e7c
|
2007-03-30T00:21:34
|
|
Tests for module 'ldexpl'.
|