On Mon, Nov 21, 2011 at 11:43:45AM -0600, Bruce Dubbs wrote:
> Ken Moffat wrote:
> >
> > Fortunately, google knew about this - it's an error in the test
> > handling of daylight saving time, which has been fixed in gnulib
> > and pulled into upstream coreutils. Detected at the end of October
> > when CEST changed to CST, but not apparently reported by anyone
> > building LFS. I got the patch and changed it to apply to the
> > renamed irectory in coreutils - solved the problem but I'm not sure
> > if it is the correct fix.
> >
> > Either nobody who has changed from summertime to wintertime is
> > running the tests for coreutils, or tmy build environment is
> > defective. My guess is the latter - printenv in chroot shows no
> > locale variables. AFAIK this is correct (they get cleared by
> > env -i in section 6.4 when we enter chroot), but I wonder if I've
> > missed something ?
>
> I haven't done an LFS build since the time change. Can you post the patch?
>
Sure - I thought about committing it, but I'd rather wait to find
out if it is needed, or if is highlighting an environmental problem.
Earlier parts of the test program spewed out variosu locale names
when I defined DEBUG, although LOCALE_ZH_CN was shown as 'none'.
I've installed all the locales, hopefully that doesn't make a
difference to this particular part of the test.
Better mention the test, which is why I suspect an environmental
problem:
/* ISO 8601 extended date and time of day representation,
'T' separator, local time zone */
p = "2011-05-01T11:55:18";
expected.tv_sec = 1304250918 - gmtoff;
expected.tv_nsec = 0;
ASSERT (parse_datetime (&result, p, 0));
LOG (p, expected, result);
ASSERT (expected.tv_sec == result.tv_sec
&& expected.tv_nsec == result.tv_nsec);
It was the second ASSERT that failed.
the patch:
not yet Submitted By: Ken Moffat <ken at linuxfromscratch dot org>
Date: 2011-11-21
Initial Package Version: 8.14
Upstream Status: Applied
Origin: http://thread.gmane.org/gmane.comp.lib.gnulib.bugs/28825
Description: Avoid false positive test in parse-datetime-test.
Fixed up to apply directly to coreutils instead of gnulib.
[ i.e. gnulib-tests instead of tests ]
diff --git a/tests/test-parse-datetime.c b/tests/test-parse-datetime.c
index b9d08a6..22fe9bc 100644
--- a/gnulib-tests/test-parse-datetime.c
+++ b/gnulib-tests/test-parse-datetime.c
@@ -94,20 +94,17 @@ tm_diff (struct tm const *a, struct tm const *b)
#endif /* ! HAVE_TM_GMTOFF */
static long
-gmt_offset ()
+gmt_offset (time_t s)
{
- time_t now;
long gmtoff;
- time (&now);
-
#if !HAVE_TM_GMTOFF
- struct tm tm_local = *localtime (&now);
- struct tm tm_gmt = *gmtime (&now);
+ struct tm tm_local = *localtime (&s);
+ struct tm tm_gmt = *gmtime (&s);
gmtoff = tm_diff (&tm_local, &tm_gmt);
#else
- gmtoff = localtime (&now)->tm_gmtoff;
+ gmtoff = localtime (&s)->tm_gmtoff;
#endif
return gmtoff;
@@ -123,16 +120,17 @@ main (int argc _GL_UNUSED, char **argv)
const char *p;
int i;
long gmtoff;
+ time_t ref_time = 1304250918;
set_program_name (argv[0]);
- gmtoff = gmt_offset ();
+ gmtoff = gmt_offset (ref_time);
/* ISO 8601 extended date and time of day representation,
'T' separator, local time zone */
p = "2011-05-01T11:55:18";
- expected.tv_sec = 1304250918 - gmtoff;
+ expected.tv_sec = ref_time - gmtoff;
expected.tv_nsec = 0;
ASSERT (parse_datetime (&result, p, 0));
LOG (p, expected, result);
@@ -142,7 +140,7 @@ main (int argc _GL_UNUSED, char **argv)
/* ISO 8601 extended date and time of day representation,
' ' separator, local time zone */
p = "2011-05-01 11:55:18";
- expected.tv_sec = 1304250918 - gmtoff;
+ expected.tv_sec = ref_time - gmtoff;
expected.tv_nsec = 0;
ASSERT (parse_datetime (&result, p, 0));
LOG (p, expected, result);
@@ -153,7 +151,7 @@ main (int argc _GL_UNUSED, char **argv)
/* ISO 8601, extended date and time of day representation,
'T' separator, UTC */
p = "2011-05-01T11:55:18Z";
- expected.tv_sec = 1304250918;
+ expected.tv_sec = ref_time;
expected.tv_nsec = 0;
ASSERT (parse_datetime (&result, p, 0));
LOG (p, expected, result);
@@ -163,7 +161,7 @@ main (int argc _GL_UNUSED, char **argv)
/* ISO 8601, extended date and time of day representation,
' ' separator, UTC */
p = "2011-05-01 11:55:18Z";
- expected.tv_sec = 1304250918;
+ expected.tv_sec = ref_time;
expected.tv_nsec = 0;
ASSERT (parse_datetime (&result, p, 0));
LOG (p, expected, result);
--
das eine Mal als Tragödie, das andere Mal als Farce
--
http://linuxfromscratch.org/mailman/listinfo/lfs-dev
FAQ: http://www.linuxfromscratch.org/faq/
Unsubscribe: See the above information page
