Re: mkstemp always generates the same file if removed between invocations

2023-10-13 Thread Bruno Haible
Hi, Sterpu Mihai wrote: > It seems the underlying implementation of mskstemp is different between > glibc and gnulib. > Specifically, in tempname.c function try_tempname_len. > The local variable v of type random_value is initialized to 0 in gnulib: > > /* A random variable. */ > random_valu

tests: Make use of signed signalling NaNs

2023-10-13 Thread Bruno Haible
Just like we have a module for signed quiet NaNs, let me introduce a module for signed signalling NaNs. So that we can test signed signalling NaNs as arguments of various functions. 2023-10-13 Bruno Haible tests: Make use of signed signalling NaNs. * tests/test-signbit.c: Incl

Re: new modules nan, qnan, snan

2023-10-13 Thread Bruno Haible
Hi Pádraig, > coreutils ci fails due to modules/snan referring to > a non existent m4/snan.m4 file. Oops. Thanks for the notice. Just fixed that. Bruno

Re: new modules nan, qnan, snan

2023-10-13 Thread Pádraig Brady
Hi Bruno, coreutils ci fails due to modules/snan referring to a non existent m4/snan.m4 file. cheers, Pádraig

Re: access: Work around trailing slash bug on Mac OS X 10.5

2023-10-13 Thread Bruno Haible
On 2023-10-03 I did: > * tests/test-access.c (main): Test for result if the argument has a > trailing slash. The new test cases fail on native Windows: ../../gltests/test-access.h:54: assertion 'errno == ENOTDIR' failed FAIL test-access.exe (exit status: 3) This patch fixes it.

isnanf, isnand: Fix conflict with Solaris

2023-10-13 Thread Bruno Haible
On Solaris 10 and 11, I see a compilation error: In file included from ../../gltests/snan.h:22:0, from ../../gltests/test-snan-2.c:22: /usr/include/ieeefp.h:159:12: error: expected declaration specifiers or ‘...’ before ‘(’ token extern int isnanf(float); ^ /usr/incl