Re: gnulib support for st_birthtime (second revision of patch)

[James Youngman]

On 3/29/07, Bruno Haible <address@hidden> wrote:
> On AIX 5.1, with gnulib as of today (29 March) the test-stat-time test fails.
> More precisely, on a 'jfs2' file system, the test fails in the third ASSERT
> of test_mtime. I get these time stamps:
[...]
> I can see two ways to fix the test:
> - Turn this ASSERT into an 'if' that does appropriate action if not fulfilled.
> - Immediately after renaming the file, write something into it, so that also
>   on this weird filesystem the ctime is bumped.

Does the following patch fix the problem?

2007-03-30  James Youngman  <address@hidden>

        * tests/test-stat-time.c (prepare_test): use chmod() rather than
        rename() to change the ctime of a file (because ctime is
        unaffected by rename on jfs2 on AIX 5.1).
        (main): Start by doing cleanup, in case a previous run failed
        leaving test files behind.


Index: test-stat-time.c
===================================================================
RCS file: /sources/gnulib/gnulib/tests/test-stat-time.c,v
retrieving revision 1.3
diff -u -p -r1.3 test-stat-time.c
--- test-stat-time.c    27 Mar 2007 18:51:18 -0000      1.3
+++ test-stat-time.c    29 Mar 2007 23:22:18 -0000
@@ -84,12 +84,12 @@ prepare_test (struct stat *statinfo, str
  sleep (2);
  create_file ("t-stt-stamp2");
  sleep (2);
-  ASSERT (rename ("t-stt-testfile", "t-stt-renamed") == 0);
+  ASSERT (chmod ("t-stt-testfile", 0400) == 0);
  sleep (2);
  create_file ("t-stt-stamp3");

  do_stat ("t-stt-stamp1",  &statinfo[0]);
-  do_stat ("t-stt-renamed", &statinfo[1]);
+  do_stat ("t-stt-testfile", &statinfo[1]);
  do_stat ("t-stt-stamp2",  &statinfo[2]);
  do_stat ("t-stt-stamp3",  &statinfo[3]);

@@ -164,6 +164,7 @@ main ()
  signal (SIGTERM, cleanup);
#endif

+  cleanup (0);
  prepare_test (statinfo, modtimes);
  test_mtime (statinfo, modtimes);
  test_birthtime (statinfo, modtimes, birthtimes);