Re: Failing Autotests / Bugs

Am 21.08.2011 16:21, schrieb Daniel Convissor: > Hi Reindl: > > On Sun, Aug 21, 2011 at 03:08:09PM +0200, Reindl Harald wrote: >> This is my try to help badly prevent broken releases like >> 5.3.7 in the future, 54 failing tests are way too much and >> the XFAILED should be removed until they are working > XFAIL's should stay, in my opinion. Their results go to the "EXPECTED > FAILED TEST SUMMARY" so there's no confusion with stuff in the "FAILED > TEST SUMMARY." not really because most of them are not working ones and they should not appear as long they are working - in my opinion a test have to return PASSED and never "expected to fail" >> FAIL DateTime::diff() days -- spring type2 type2 >> [ext/date/tests/DateTime_days-spring-type2-type2.phpt] >> FAIL DateTime::diff() days -- spring type2 type3 >> [ext/date/tests/DateTime_days-spring-type2-type3.phpt] >> FAIL DateTime::diff() days -- spring type3 type2 >> [ext/date/tests/DateTime_days-spring-type3-type2.phpt] >> FAIL DateTime::diff() days -- spring type3 type3 >> [ext/date/tests/DateTime_days-spring-type3-type3.phpt] >> FAIL DateTime::sub() -- dates [ext/date/tests/DateTime_sub-dates.phpt] >> FAIL DateTime::sub() -- february [ext/date/tests/DateTime_sub-february.phpt] > That's weird. Those are marked for XFAIL. How were you running the > tests and on what checkout/code base? via rpmbuild / SPEC-File based on the fedora one %check %if %runselftest cd build-apache # Run tests, using the CLI SAPI export NO_INTERACTION=1 REPORT_EXIT_STATUS=1 MALLOC_CHECK_=2 MYSQL_TEST_HOST="localhost" MYSQL_TEST_PORT="3306" MYSQL_TEST_USER="php_autotest" MYSQL_TEST_PASSWD="***********" MYSQL_TEST_DB="php_autotest" unset TZ LANG LC_ALL if ! make test; then set +x for f in find .. -name \*.diff -type f -print; do echo "TEST FAILURE: $f --" cat "$f" echo "-- $f result ends." done set -x #exit 1 fi unset NO_INTERACTION REPORT_EXIT_STATUS MALLOC_CHECK_ %endif > Also, it looks like you're reading from the main output list, not the > various SUMMARY sections, which is totally inefficient because it lists > passes, skips, etc, too. it is not because grep was invited

Attachment: [application/pgp-signature] OpenPGP digital signature signature.asc