On 08/12/2018 01:19 AM, Xi Ruoyao wrote:
On 2018-08-11 18:07 +0100, Ken Moffat wrote:
I'm running all the tests (except for vim - I think that dislikes my
use of urxvt) and I wasn't intending to look at the results until
later (this build is experimental, to see how much, if any, damage
is caused by various potential BLFS updates). But for grep 'make
check' unexpectedly failed (status 1) after an XFAIL.
XFAIL triple-backref (exit status: 1)
It's a good thing. Glibc-2.28 fixed a bug so this test should not
FAIL now. See thread
<http://lists.gnu.org/archive/html/bug-grep/2018-08/msg00000.html>.
The voluminous output in test-suite.log seems to be the same as
what I logged from stdout and stderr, and looking closely I now
notice:
grep: regexec.c:1359: pop_fail_stack: Assertion `num >= 0' failed.
./triple-backref: line 15: 8025 Aborted (core dumped) grep -E
'(.?)(.?)(.?)\3\2\1' in > out
+ fail=1
Looking at the log from the same version (3.1) when I last built it,
the main tests had two XFAIL, backref-alt and triple-backref but
WITHOUT an assertion failure.
I rebuilt grep-3.1 on my LFS-8.2 with Glibc-2.27. Both
triple-backref and backref-alt aborted because of fail assertions.
Please recheck your result.
make[3]: Entering directory '/building/grep-3.1/tests'
============================================================================
Testsuite summary for GNU grep 3.1
============================================================================
# TOTAL: 109
# PASS: 88
# SKIP: 19
# XFAIL: 2
# FAIL: 0
# XPASS: 0
# ERROR: 0
============================================================================
make[3]: Leaving directory '/building/grep-3.1/tests'
Making check in gnulib-tests
and in gnulib-tests on that build 134 tests passed and 7 were
skipped, but on today's build gnulib is not mentioned in the test
output.
The "main" test "failed" (XPASS in fact) then `make` just bailed out.
With "make check -k" gnulib-test would run.
We can
sed '/^am__append_2/d' -i tests/Makefile
after configuring Grep, or just use "make check -k" and tell the
audiences there would be a XPASS test with Glibc-2.28.
We have not mentioned XPASS (or XFAIL or unsupported) results in the
past. I'd prefer not to set the precedent.
No problem with passing -k. I'll commit that later today.
-- Bruce
--
http://lists.linuxfromscratch.org/listinfo/lfs-dev
FAQ: http://www.linuxfromscratch.org/faq/
Unsubscribe: See the above information page
