Re: [PATCH] strerror: enforce POSIX ruling on strerror(0)

[Bruno Haible]

Eric Blake wrote on 2011-05-19:
> http://austingroupbugs.net/view.php?id=382 requires that strerror(0)
> succeed, but FreeBSD reports "Unknown error: 0" and fails with EINVAL.
> 
> * m4/strerror.m4 (gl_FUNC_STRERROR_SEPARATE): Expose BSD bug.
> * m4/strerror_r.m4 (gl_FUNC_STRERROR_R): Likewise.
> * lib/strerror_r.c (rpl_strerror_r): Work around it.
> * doc/posix-functions/strerror.texi (strerror): Document it.
> * doc/posix-functions/strerror_r.texi (strerror_r): Likewise.
> * tests/test-strerror.c (main): Strengthen test.
> * tests/test-strerror_r.c (main): Likewise.

You have marked this bug as "fixed by gnulib", but I'm still seeing test
failures about it on FreeBSD 6.4. A testdir for strerror, strerror_r-posix,
perror yields these test failures:

  test-perror2.c:112: assertion failed
  Abort trap (core dumped)
  FAIL: test-perror2

  test-strerror.c:62: assertion failed
  Abort trap (core dumped)
  FAIL: test-strerror

Relevant configure output:

checking whether strerror_r is declared... yes
checking for strerror_r... yes
checking for strerror_r with POSIX signature... yes
checking whether strerror_r works... no                        ; exit status 20
checking whether strerror_r is declared without a macro... yes
checking for working strerror function... no                   ; exit status 4
checking whether strerror is declared... yes

Relevant config.status variables:

S["REPLACE_STRERROR_R"]="1"
S["REPLACE_STRERROR"]="1"
S["HAVE_DECL_STRERROR_R"]="1"
S["GNULIB_STRERROR_R"]="1"
S["GNULIB_STRERROR"]="1"
S["REPLACE_PERROR"]="1"
S["GNULIB_PERROR"]="1"
D["HAVE_DECL_STRERROR_R"]=" 1"
D["HAVE_STRERROR_R"]=" 1"
D["HAVE_RAW_DECL_STRERROR_R"]=" 1"
D["GNULIB_TEST_PERROR"]=" 1"
D["HAVE_DECL_STRERROR"]=" 1"
D["GNULIB_STRERROR"]=" 1"
D["GNULIB_TEST_STRERROR"]=" 1"
D["HAVE_STRERROR_R"]=" 1"
D["GNULIB_TEST_STRERROR_R"]=" 1"

Bruno
-- 
In memoriam Mordechai Gebirtig <http://en.wikipedia.org/wiki/Mordechai_Gebirtig>