sin/samba-mirror
1
0
Fork 0
mirror of https://github.com/samba-team/samba.git synced 2025-01-07 17:18:11 +03:00
Code
068d4336d6
samba-mirror/source4/selftest/config.mk

122 lines
5.0 KiB
Makefile
Raw Normal View History

r26546: Rather than having a gazillion targets, add output format variable for 'make test'. (This used to be commit bd5217906c12c36df9769fc32aee77d9c1805bf2)
2007-12-20 17:54:00 +03:00
TEST_FORMAT = plain
s4:selftest Fix 'make testenv-gdb' and 'make valgrindtest' etc The problem here was that tridge's changes to ensure that test results were always propogated didn't merge well with the addition of extra environment variables for the 'make valgrindtest' and similar environments. By splitting out the macro further, we keep the build farm reporting accurate, but allow these other test modes to work. Andrew Bartlett
2009-09-08 04:48:40 +04:00
SELFTEST = $(LD_LIBPATH_OVERRIDE) PYTHON=$(PYTHON) \
selftest: pass down the path to python for provisioning samba4 metze
2009-01-30 22:38:27 +03:00
$(PERL) $(selftestdir)/selftest.pl --prefix=${selftest_prefix} \
r25707: Move selftest makefile code to separate file. (This used to be commit 9714db7814fd43d2a10485c041bbb761360c0826)
2007-10-22 03:05:31 +04:00
--builddir=$(builddir) --srcdir=$(srcdir) \
Move selftest code to top-level.
2008-09-18 21:51:03 +04:00
--exclude=$(srcdir)/selftest/skip --testlist="./selftest/tests.sh|" \
fixed make test reporting success when provision fails We were getting build test 'pass' on the build farm when the provision is failing. The cause was the use of filter pipelines in the selftest code in the makefile. A bourne shell will by default report the error status of the last element of the pipeline, which means when provision failed, make thought everything had passed. We could use the bash 'pipefail' option to fix this, but it isn't portable, so instead I've modified the makefile to touch st/st_done when the first element in the pipeline completes successfully. That means we now get errors in make test, at the expense of making the makefile even more horrible than it was.
2009-08-17 16:45:51 +04:00
$(TEST_OPTIONS)
# this strange code is to handle the fact that the bash pipefail option is
# not portable. When we run selftest we use a pipeline, and the first element
# of that pipeline might abort. We need to catch this and mark the test
# as having failed
s4:selftest Fix 'make testenv-gdb' and 'make valgrindtest' etc The problem here was that tridge's changes to ensure that test results were always propogated didn't merge well with the addition of extra environment variables for the 'make valgrindtest' and similar environments. By splitting out the macro further, we keep the build farm reporting accurate, but allow these other test modes to work. Andrew Bartlett
2009-09-08 04:48:40 +04:00
ST_RM = ( rm -f $(selftest_prefix)/st_done &&
fixed make test reporting success when provision fails We were getting build test 'pass' on the build farm when the provision is failing. The cause was the use of filter pipelines in the selftest code in the makefile. A bourne shell will by default report the error status of the last element of the pipeline, which means when provision failed, make thought everything had passed. We could use the bash 'pipefail' option to fix this, but it isn't portable, so instead I've modified the makefile to touch st/st_done when the first element in the pipeline completes successfully. That means we now get errors in make test, at the expense of making the makefile even more horrible than it was.
2009-08-17 16:45:51 +04:00
ST_TOUCH = && touch $(selftest_prefix)/st_done )
ST_DONE_TEST = @test -f $(selftest_prefix)/st_done || { echo "SELFTEST FAILED"; exit 1; }
r25707: Move selftest makefile code to separate file. (This used to be commit 9714db7814fd43d2a10485c041bbb761360c0826)
2007-10-22 03:05:31 +04:00
Move selftest code to top-level.
2008-09-18 21:51:03 +04:00
SELFTEST_NOSLOW_OPTS = --exclude=$(srcdir)/selftest/slow
Require separate option for quicktest include list rather than making assumptions about it.
2008-10-22 06:09:17 +04:00
SELFTEST_QUICK_OPTS = $(SELFTEST_NOSLOW_OPTS) --quick --include=$(srcdir)/selftest/quick
selftest: Fix subunit stream to include the right prefixes rather than extending the subunit protocol.
2009-06-05 15:16:46 +04:00
FILTER_XFAIL = $(PERL) $(selftestdir)/filter-subunit.pl --expected-failures=$(srcdir)/selftest/knownfail
s4:selftest: place summary file into selftest-prefix metze
2009-11-16 18:48:18 +03:00
FORMAT_TEST = $(PERL) $(selftestdir)/format-subunit.pl --prefix=${selftest_prefix} --format=$(TEST_FORMAT)
FORMAT_TEST_OUTPUT = $(FILTER_XFAIL) | $(FORMAT_TEST)
selftest: Use filter-xfail for known failures
2009-06-03 20:19:01 +04:00
selftest: Fix handling of testsuite, reintroduce progress indication.
2009-06-05 18:10:12 +04:00
test-subunit:: everything
fixed make test reporting success when provision fails We were getting build test 'pass' on the build farm when the provision is failing. The cause was the use of filter pipelines in the selftest code in the makefile. A bourne shell will by default report the error status of the last element of the pipeline, which means when provision failed, make thought everything had passed. We could use the bash 'pipefail' option to fix this, but it isn't portable, so instead I've modified the makefile to touch st/st_done when the first element in the pipeline completes successfully. That means we now get errors in make test, at the expense of making the makefile even more horrible than it was.
2009-08-17 16:45:51 +04:00
$(SELFTEST) --socket-wrapper $(TESTS) $(ST_TOUCH)
$(ST_DONE_TEST)
Add samba4-slow file listing tests that succeed but are just slow. Added a 'make slowtest' target that does run these tests. Also added some comments in the test lists files. (This used to be commit f67f3422b0fd319a3c6d3ccb3ea444a97c3ebaae)
2008-05-29 20:19:19 +04:00
slowtest:: everything
fixed make test reporting success when provision fails We were getting build test 'pass' on the build farm when the provision is failing. The cause was the use of filter pipelines in the selftest code in the makefile. A bourne shell will by default report the error status of the last element of the pipeline, which means when provision failed, make thought everything had passed. We could use the bash 'pipefail' option to fix this, but it isn't portable, so instead I've modified the makefile to touch st/st_done when the first element in the pipeline completes successfully. That means we now get errors in make test, at the expense of making the makefile even more horrible than it was.
2009-08-17 16:45:51 +04:00
$(SELFTEST) $(DEFAULT_TEST_OPTIONS) $(TESTS) $(ST_TOUCH) | $(FORMAT_TEST_OUTPUT) --immediate
$(ST_DONE_TEST)
r25707: Move selftest makefile code to separate file. (This used to be commit 9714db7814fd43d2a10485c041bbb761360c0826)
2007-10-22 03:05:31 +04:00
selftest: Output subunit on the buildfarm.
2009-06-12 16:21:03 +04:00
ifeq ($(RUN_FROM_BUILD_FARM),yes)
test:: everything
s4:selftest Fix 'make testenv-gdb' and 'make valgrindtest' etc The problem here was that tridge's changes to ensure that test results were always propogated didn't merge well with the addition of extra environment variables for the 'make valgrindtest' and similar environments. By splitting out the macro further, we keep the build farm reporting accurate, but allow these other test modes to work. Andrew Bartlett
2009-09-08 04:48:40 +04:00
$(ST_RM) $(SELFTEST) $(SELFTEST_NOSLOW_OPTS) $(DEFAULT_TEST_OPTIONS) $(TESTS) $(ST_TOUCH) | $(FILTER_XFAIL) --strip-passed-output
fixed make test reporting success when provision fails We were getting build test 'pass' on the build farm when the provision is failing. The cause was the use of filter pipelines in the selftest code in the makefile. A bourne shell will by default report the error status of the last element of the pipeline, which means when provision failed, make thought everything had passed. We could use the bash 'pipefail' option to fix this, but it isn't portable, so instead I've modified the makefile to touch st/st_done when the first element in the pipeline completes successfully. That means we now get errors in make test, at the expense of making the makefile even more horrible than it was.
2009-08-17 16:45:51 +04:00
$(ST_DONE_TEST)
selftest: Output subunit on the buildfarm.
2009-06-12 16:21:03 +04:00
else
Add samba4-slow file listing tests that succeed but are just slow. Added a 'make slowtest' target that does run these tests. Also added some comments in the test lists files. (This used to be commit f67f3422b0fd319a3c6d3ccb3ea444a97c3ebaae)
2008-05-29 20:19:19 +04:00
test:: everything
s4:selftest Fix 'make testenv-gdb' and 'make valgrindtest' etc The problem here was that tridge's changes to ensure that test results were always propogated didn't merge well with the addition of extra environment variables for the 'make valgrindtest' and similar environments. By splitting out the macro further, we keep the build farm reporting accurate, but allow these other test modes to work. Andrew Bartlett
2009-09-08 04:48:40 +04:00
$(ST_RM) $(SELFTEST) $(SELFTEST_NOSLOW_OPTS) $(DEFAULT_TEST_OPTIONS) $(TESTS) $(ST_TOUCH) | $(FORMAT_TEST_OUTPUT) --immediate
fixed make test reporting success when provision fails We were getting build test 'pass' on the build farm when the provision is failing. The cause was the use of filter pipelines in the selftest code in the makefile. A bourne shell will by default report the error status of the last element of the pipeline, which means when provision failed, make thought everything had passed. We could use the bash 'pipefail' option to fix this, but it isn't portable, so instead I've modified the makefile to touch st/st_done when the first element in the pipeline completes successfully. That means we now get errors in make test, at the expense of making the makefile even more horrible than it was.
2009-08-17 16:45:51 +04:00
$(ST_DONE_TEST)
selftest: Output subunit on the buildfarm.
2009-06-12 16:21:03 +04:00
endif
Add samba4-slow file listing tests that succeed but are just slow. Added a 'make slowtest' target that does run these tests. Also added some comments in the test lists files. (This used to be commit f67f3422b0fd319a3c6d3ccb3ea444a97c3ebaae)
2008-05-29 20:19:19 +04:00
Add Kvm start code. (This used to be commit 6308c134b225dfa8574aa1370a7874ed6e572ccc)
2008-02-19 05:29:03 +03:00
kvmtest:: everything
s4:selftest Fix 'make testenv-gdb' and 'make valgrindtest' etc The problem here was that tridge's changes to ensure that test results were always propogated didn't merge well with the addition of extra environment variables for the 'make valgrindtest' and similar environments. By splitting out the macro further, we keep the build farm reporting accurate, but allow these other test modes to work. Andrew Bartlett
2009-09-08 04:48:40 +04:00
$(ST_RM) $(SELFTEST) $(SELFTEST_NOSLOW_OPTS) $(DEFAULT_TEST_OPTIONS) \
fixed make test reporting success when provision fails We were getting build test 'pass' on the build farm when the provision is failing. The cause was the use of filter pipelines in the selftest code in the makefile. A bourne shell will by default report the error status of the last element of the pipeline, which means when provision failed, make thought everything had passed. We could use the bash 'pipefail' option to fix this, but it isn't portable, so instead I've modified the makefile to touch st/st_done when the first element in the pipeline completes successfully. That means we now get errors in make test, at the expense of making the makefile even more horrible than it was.
2009-08-17 16:45:51 +04:00
--target=kvm --image=$(KVM_IMAGE) $(ST_TOUCH) | $(FORMAT_TEST_OUTPUT) --immediate
$(ST_DONE_TEST)
Add Kvm start code. (This used to be commit 6308c134b225dfa8574aa1370a7874ed6e572ccc)
2008-02-19 05:29:03 +03:00
Start and kill kvm instance properly. (This used to be commit 78e49765a2ec5fac485dbf56143716e151b4b562)
2008-02-19 17:12:57 +03:00
kvmquicktest:: everything
s4:selftest Fix 'make testenv-gdb' and 'make valgrindtest' etc The problem here was that tridge's changes to ensure that test results were always propogated didn't merge well with the addition of extra environment variables for the 'make valgrindtest' and similar environments. By splitting out the macro further, we keep the build farm reporting accurate, but allow these other test modes to work. Andrew Bartlett
2009-09-08 04:48:40 +04:00
$(ST_RM) $(SELFTEST) $(DEFAULT_TEST_OPTIONS) \
fixed make test reporting success when provision fails We were getting build test 'pass' on the build farm when the provision is failing. The cause was the use of filter pipelines in the selftest code in the makefile. A bourne shell will by default report the error status of the last element of the pipeline, which means when provision failed, make thought everything had passed. We could use the bash 'pipefail' option to fix this, but it isn't portable, so instead I've modified the makefile to touch st/st_done when the first element in the pipeline completes successfully. That means we now get errors in make test, at the expense of making the makefile even more horrible than it was.
2009-08-17 16:45:51 +04:00
$(SELFTEST_QUICK_OPTS) --target=kvm --image=$(KVM_IMAGE) $(ST_TOUCH) | $(FORMAT_TEST_OUTPUT) | $(FORMAT_TEST_OUTPUT) --immediate
$(ST_DONE_TEST)
Start and kill kvm instance properly. (This used to be commit 78e49765a2ec5fac485dbf56143716e151b4b562)
2008-02-19 17:12:57 +03:00
Use mkinclude (makes the Makefile a bit shorter), use double-colon for phony makefile targets. (This used to be commit ea7d389bcc284d2e9eeff79bce5c285269099ad3)
2008-02-09 18:05:55 +03:00
testone:: everything
s4:selftest Fix 'make testenv-gdb' and 'make valgrindtest' etc The problem here was that tridge's changes to ensure that test results were always propogated didn't merge well with the addition of extra environment variables for the 'make valgrindtest' and similar environments. By splitting out the macro further, we keep the build farm reporting accurate, but allow these other test modes to work. Andrew Bartlett
2009-09-08 04:48:40 +04:00
$(ST_RM) $(SELFTEST) $(SELFTEST_NOSLOW_OPTS) $(DEFAULT_TEST_OPTIONS) --one $(TESTS) $(ST_TOUCH) | $(FORMAT_TEST_OUTPUT)
fixed make test reporting success when provision fails We were getting build test 'pass' on the build farm when the provision is failing. The cause was the use of filter pipelines in the selftest code in the makefile. A bourne shell will by default report the error status of the last element of the pipeline, which means when provision failed, make thought everything had passed. We could use the bash 'pipefail' option to fix this, but it isn't portable, so instead I've modified the makefile to touch st/st_done when the first element in the pipeline completes successfully. That means we now get errors in make test, at the expense of making the makefile even more horrible than it was.
2009-08-17 16:45:51 +04:00
$(ST_DONE_TEST)
r25707: Move selftest makefile code to separate file. (This used to be commit 9714db7814fd43d2a10485c041bbb761360c0826)
2007-10-22 03:05:31 +04:00
Use mkinclude (makes the Makefile a bit shorter), use double-colon for phony makefile targets. (This used to be commit ea7d389bcc284d2e9eeff79bce5c285269099ad3)
2008-02-09 18:05:55 +03:00
test-swrap:: everything
s4:selftest Fix 'make testenv-gdb' and 'make valgrindtest' etc The problem here was that tridge's changes to ensure that test results were always propogated didn't merge well with the addition of extra environment variables for the 'make valgrindtest' and similar environments. By splitting out the macro further, we keep the build farm reporting accurate, but allow these other test modes to work. Andrew Bartlett
2009-09-08 04:48:40 +04:00
$(ST_RM) $(SELFTEST) $(SELFTEST_NOSLOW_OPTS) --socket-wrapper $(TESTS) $(ST_TOUCH) | $(FORMAT_TEST_OUTPUT) --immediate
fixed make test reporting success when provision fails We were getting build test 'pass' on the build farm when the provision is failing. The cause was the use of filter pipelines in the selftest code in the makefile. A bourne shell will by default report the error status of the last element of the pipeline, which means when provision failed, make thought everything had passed. We could use the bash 'pipefail' option to fix this, but it isn't portable, so instead I've modified the makefile to touch st/st_done when the first element in the pipeline completes successfully. That means we now get errors in make test, at the expense of making the makefile even more horrible than it was.
2009-08-17 16:45:51 +04:00
$(ST_DONE_TEST)
r25707: Move selftest makefile code to separate file. (This used to be commit 9714db7814fd43d2a10485c041bbb761360c0826)
2007-10-22 03:05:31 +04:00
Use mkinclude (makes the Makefile a bit shorter), use double-colon for phony makefile targets. (This used to be commit ea7d389bcc284d2e9eeff79bce5c285269099ad3)
2008-02-09 18:05:55 +03:00
test-swrap-pcap:: everything
s4:selftest Fix 'make testenv-gdb' and 'make valgrindtest' etc The problem here was that tridge's changes to ensure that test results were always propogated didn't merge well with the addition of extra environment variables for the 'make valgrindtest' and similar environments. By splitting out the macro further, we keep the build farm reporting accurate, but allow these other test modes to work. Andrew Bartlett
2009-09-08 04:48:40 +04:00
$(ST_RM) $(SELFTEST) $(SELFTEST_NOSLOW_OPTS) --socket-wrapper-pcap $(TESTS) $(ST_TOUCH) | $(FORMAT_TEST_OUTPUT) --immediate
fixed make test reporting success when provision fails We were getting build test 'pass' on the build farm when the provision is failing. The cause was the use of filter pipelines in the selftest code in the makefile. A bourne shell will by default report the error status of the last element of the pipeline, which means when provision failed, make thought everything had passed. We could use the bash 'pipefail' option to fix this, but it isn't portable, so instead I've modified the makefile to touch st/st_done when the first element in the pipeline completes successfully. That means we now get errors in make test, at the expense of making the makefile even more horrible than it was.
2009-08-17 16:45:51 +04:00
$(ST_DONE_TEST)
r25707: Move selftest makefile code to separate file. (This used to be commit 9714db7814fd43d2a10485c041bbb761360c0826)
2007-10-22 03:05:31 +04:00
Use mkinclude (makes the Makefile a bit shorter), use double-colon for phony makefile targets. (This used to be commit ea7d389bcc284d2e9eeff79bce5c285269099ad3)
2008-02-09 18:05:55 +03:00
test-swrap-keep-pcap:: everything
s4:selftest Fix 'make testenv-gdb' and 'make valgrindtest' etc The problem here was that tridge's changes to ensure that test results were always propogated didn't merge well with the addition of extra environment variables for the 'make valgrindtest' and similar environments. By splitting out the macro further, we keep the build farm reporting accurate, but allow these other test modes to work. Andrew Bartlett
2009-09-08 04:48:40 +04:00
$(ST_RM) $(SELFTEST) $(SELFTEST_NOSLOW_OPTS) --socket-wrapper-keep-pcap $(TESTS) $(ST_TOUCH) | $(FORMAT_TEST_OUTPUT) --immediate
fixed make test reporting success when provision fails We were getting build test 'pass' on the build farm when the provision is failing. The cause was the use of filter pipelines in the selftest code in the makefile. A bourne shell will by default report the error status of the last element of the pipeline, which means when provision failed, make thought everything had passed. We could use the bash 'pipefail' option to fix this, but it isn't portable, so instead I've modified the makefile to touch st/st_done when the first element in the pipeline completes successfully. That means we now get errors in make test, at the expense of making the makefile even more horrible than it was.
2009-08-17 16:45:51 +04:00
$(ST_DONE_TEST)
r25707: Move selftest makefile code to separate file. (This used to be commit 9714db7814fd43d2a10485c041bbb761360c0826)
2007-10-22 03:05:31 +04:00
Use mkinclude (makes the Makefile a bit shorter), use double-colon for phony makefile targets. (This used to be commit ea7d389bcc284d2e9eeff79bce5c285269099ad3)
2008-02-09 18:05:55 +03:00
test-noswrap:: everything
s4:selftest Fix 'make testenv-gdb' and 'make valgrindtest' etc The problem here was that tridge's changes to ensure that test results were always propogated didn't merge well with the addition of extra environment variables for the 'make valgrindtest' and similar environments. By splitting out the macro further, we keep the build farm reporting accurate, but allow these other test modes to work. Andrew Bartlett
2009-09-08 04:48:40 +04:00
$(ST_RM) $(SELFTEST) $(SELFTEST_NOSLOW_OPTS) $(TESTS) $(ST_TOUCH) | $(FORMAT_TEST_OUTPUT) --immediate
fixed make test reporting success when provision fails We were getting build test 'pass' on the build farm when the provision is failing. The cause was the use of filter pipelines in the selftest code in the makefile. A bourne shell will by default report the error status of the last element of the pipeline, which means when provision failed, make thought everything had passed. We could use the bash 'pipefail' option to fix this, but it isn't portable, so instead I've modified the makefile to touch st/st_done when the first element in the pipeline completes successfully. That means we now get errors in make test, at the expense of making the makefile even more horrible than it was.
2009-08-17 16:45:51 +04:00
$(ST_DONE_TEST)
r25707: Move selftest makefile code to separate file. (This used to be commit 9714db7814fd43d2a10485c041bbb761360c0826)
2007-10-22 03:05:31 +04:00
Use mkinclude (makes the Makefile a bit shorter), use double-colon for phony makefile targets. (This used to be commit ea7d389bcc284d2e9eeff79bce5c285269099ad3)
2008-02-09 18:05:55 +03:00
quicktest:: all
s4:selftest Fix 'make testenv-gdb' and 'make valgrindtest' etc The problem here was that tridge's changes to ensure that test results were always propogated didn't merge well with the addition of extra environment variables for the 'make valgrindtest' and similar environments. By splitting out the macro further, we keep the build farm reporting accurate, but allow these other test modes to work. Andrew Bartlett
2009-09-08 04:48:40 +04:00
$(ST_RM) $(SELFTEST) $(SELFTEST_QUICK_OPTS) --socket-wrapper $(TESTS) $(ST_TOUCH) | $(FORMAT_TEST_OUTPUT) --immediate
fixed make test reporting success when provision fails We were getting build test 'pass' on the build farm when the provision is failing. The cause was the use of filter pipelines in the selftest code in the makefile. A bourne shell will by default report the error status of the last element of the pipeline, which means when provision failed, make thought everything had passed. We could use the bash 'pipefail' option to fix this, but it isn't portable, so instead I've modified the makefile to touch st/st_done when the first element in the pipeline completes successfully. That means we now get errors in make test, at the expense of making the makefile even more horrible than it was.
2009-08-17 16:45:51 +04:00
$(ST_DONE_TEST)
r25707: Move selftest makefile code to separate file. (This used to be commit 9714db7814fd43d2a10485c041bbb761360c0826)
2007-10-22 03:05:31 +04:00
selftest: Fix handling of testsuite, reintroduce progress indication.
2009-06-05 18:10:12 +04:00
quicktest-subunit:: all
s4:selftest Fix 'make testenv-gdb' and 'make valgrindtest' etc The problem here was that tridge's changes to ensure that test results were always propogated didn't merge well with the addition of extra environment variables for the 'make valgrindtest' and similar environments. By splitting out the macro further, we keep the build farm reporting accurate, but allow these other test modes to work. Andrew Bartlett
2009-09-08 04:48:40 +04:00
$(ST_RM) $(SELFTEST) $(SELFTEST_QUICK_OPTS) --socket-wrapper $(TESTS) $(ST_TOUCH)
fixed make test reporting success when provision fails We were getting build test 'pass' on the build farm when the provision is failing. The cause was the use of filter pipelines in the selftest code in the makefile. A bourne shell will by default report the error status of the last element of the pipeline, which means when provision failed, make thought everything had passed. We could use the bash 'pipefail' option to fix this, but it isn't portable, so instead I've modified the makefile to touch st/st_done when the first element in the pipeline completes successfully. That means we now get errors in make test, at the expense of making the makefile even more horrible than it was.
2009-08-17 16:45:51 +04:00
$(ST_DONE_TEST)
selftest: Fix handling of testsuite, reintroduce progress indication.
2009-06-05 18:10:12 +04:00
Use mkinclude (makes the Makefile a bit shorter), use double-colon for phony makefile targets. (This used to be commit ea7d389bcc284d2e9eeff79bce5c285269099ad3)
2008-02-09 18:05:55 +03:00
quicktestone:: all
s4:selftest Fix 'make testenv-gdb' and 'make valgrindtest' etc The problem here was that tridge's changes to ensure that test results were always propogated didn't merge well with the addition of extra environment variables for the 'make valgrindtest' and similar environments. By splitting out the macro further, we keep the build farm reporting accurate, but allow these other test modes to work. Andrew Bartlett
2009-09-08 04:48:40 +04:00
$(ST_RM) $(SELFTEST) $(SELFTEST_QUICK_OPTS) --socket-wrapper --one $(TESTS) | $(FORMAT_TEST_OUTPUT)
fixed make test reporting success when provision fails We were getting build test 'pass' on the build farm when the provision is failing. The cause was the use of filter pipelines in the selftest code in the makefile. A bourne shell will by default report the error status of the last element of the pipeline, which means when provision failed, make thought everything had passed. We could use the bash 'pipefail' option to fix this, but it isn't portable, so instead I've modified the makefile to touch st/st_done when the first element in the pipeline completes successfully. That means we now get errors in make test, at the expense of making the makefile even more horrible than it was.
2009-08-17 16:45:51 +04:00
$(ST_DONE_TEST)
r25707: Move selftest makefile code to separate file. (This used to be commit 9714db7814fd43d2a10485c041bbb761360c0826)
2007-10-22 03:05:31 +04:00
Use mkinclude (makes the Makefile a bit shorter), use double-colon for phony makefile targets. (This used to be commit ea7d389bcc284d2e9eeff79bce5c285269099ad3)
2008-02-09 18:05:55 +03:00
testenv:: everything
s4:selftest Fix 'make testenv-gdb' and 'make valgrindtest' etc The problem here was that tridge's changes to ensure that test results were always propogated didn't merge well with the addition of extra environment variables for the 'make valgrindtest' and similar environments. By splitting out the macro further, we keep the build farm reporting accurate, but allow these other test modes to work. Andrew Bartlett
2009-09-08 04:48:40 +04:00
$(SELFTEST) $(SELFTEST_NOSLOW_OPTS) --socket-wrapper --testenv
r25707: Move selftest makefile code to separate file. (This used to be commit 9714db7814fd43d2a10485c041bbb761360c0826)
2007-10-22 03:05:31 +04:00
Allow 'make testenv-dc', 'make testenv-member', etc.
2008-10-06 18:00:10 +04:00
testenv-%:: everything
s4:selftest Fix 'make testenv-gdb' and 'make valgrindtest' etc The problem here was that tridge's changes to ensure that test results were always propogated didn't merge well with the addition of extra environment variables for the 'make valgrindtest' and similar environments. By splitting out the macro further, we keep the build farm reporting accurate, but allow these other test modes to work. Andrew Bartlett
2009-09-08 04:48:40 +04:00
SELFTEST_TESTENV=$* $(SELFTEST) $(SELFTEST_NOSLOW_OPTS) --socket-wrapper --testenv
Allow 'make testenv-dc', 'make testenv-member', etc.
2008-10-06 18:00:10 +04:00
Support "make test-X" to run all tests that match regex X.
2008-10-12 00:22:29 +04:00
test-%::
$(MAKE) test TESTS=$*
Use mkinclude (makes the Makefile a bit shorter), use double-colon for phony makefile targets. (This used to be commit ea7d389bcc284d2e9eeff79bce5c285269099ad3)
2008-02-09 18:05:55 +03:00
valgrindtest:: valgrindtest-all
r25707: Move selftest makefile code to separate file. (This used to be commit 9714db7814fd43d2a10485c041bbb761360c0826)
2007-10-22 03:05:31 +04:00
Use mkinclude (makes the Makefile a bit shorter), use double-colon for phony makefile targets. (This used to be commit ea7d389bcc284d2e9eeff79bce5c285269099ad3)
2008-02-09 18:05:55 +03:00
valgrindtest-quick:: all
s4:selftest: rename SMBD_VALGRIND => SAMBA_VALGRIND metze
2009-03-20 18:11:14 +03:00
SAMBA_VALGRIND="xterm -n server -e $(selftestdir)/valgrind_run $(LD_LIBPATH_OVERRIDE)" \
r25707: Move selftest makefile code to separate file. (This used to be commit 9714db7814fd43d2a10485c041bbb761360c0826)
2007-10-22 03:05:31 +04:00
VALGRIND="valgrind -q --num-callers=30 --log-file=${selftest_prefix}/valgrind.log" \
s4:selftest Fix 'make testenv-gdb' and 'make valgrindtest' etc The problem here was that tridge's changes to ensure that test results were always propogated didn't merge well with the addition of extra environment variables for the 'make valgrindtest' and similar environments. By splitting out the macro further, we keep the build farm reporting accurate, but allow these other test modes to work. Andrew Bartlett
2009-09-08 04:48:40 +04:00
$(SELFTEST) $(SELFTEST_QUICK_OPTS) --socket-wrapper $(TESTS) | $(FORMAT_TEST_OUTPUT) --immediate
r25707: Move selftest makefile code to separate file. (This used to be commit 9714db7814fd43d2a10485c041bbb761360c0826)
2007-10-22 03:05:31 +04:00
Use mkinclude (makes the Makefile a bit shorter), use double-colon for phony makefile targets. (This used to be commit ea7d389bcc284d2e9eeff79bce5c285269099ad3)
2008-02-09 18:05:55 +03:00
valgrindtest-all:: everything
s4:selftest: rename SMBD_VALGRIND => SAMBA_VALGRIND metze
2009-03-20 18:11:14 +03:00
SAMBA_VALGRIND="xterm -n server -e $(selftestdir)/valgrind_run $(LD_LIBPATH_OVERRIDE)" \
r25707: Move selftest makefile code to separate file. (This used to be commit 9714db7814fd43d2a10485c041bbb761360c0826)
2007-10-22 03:05:31 +04:00
VALGRIND="valgrind -q --num-callers=30 --log-file=${selftest_prefix}/valgrind.log" \
s4:selftest Fix 'make testenv-gdb' and 'make valgrindtest' etc The problem here was that tridge's changes to ensure that test results were always propogated didn't merge well with the addition of extra environment variables for the 'make valgrindtest' and similar environments. By splitting out the macro further, we keep the build farm reporting accurate, but allow these other test modes to work. Andrew Bartlett
2009-09-08 04:48:40 +04:00
$(SELFTEST) $(SELFTEST_NOSLOW_OPTS) --socket-wrapper $(TESTS) | $(FORMAT_TEST_OUTPUT) --immediate
r25707: Move selftest makefile code to separate file. (This used to be commit 9714db7814fd43d2a10485c041bbb761360c0826)
2007-10-22 03:05:31 +04:00
Use mkinclude (makes the Makefile a bit shorter), use double-colon for phony makefile targets. (This used to be commit ea7d389bcc284d2e9eeff79bce5c285269099ad3)
2008-02-09 18:05:55 +03:00
valgrindtest-env:: everything
s4:selftest: rename SMBD_VALGRIND => SAMBA_VALGRIND metze
2009-03-20 18:11:14 +03:00
SAMBA_VALGRIND="xterm -n server -e $(selftestdir)/valgrind_run $(LD_LIBPATH_OVERRIDE)" \
r25707: Move selftest makefile code to separate file. (This used to be commit 9714db7814fd43d2a10485c041bbb761360c0826)
2007-10-22 03:05:31 +04:00
VALGRIND="valgrind -q --num-callers=30 --log-file=${selftest_prefix}/valgrind.log" \
s4:selftest Fix 'make testenv-gdb' and 'make valgrindtest' etc The problem here was that tridge's changes to ensure that test results were always propogated didn't merge well with the addition of extra environment variables for the 'make valgrindtest' and similar environments. By splitting out the macro further, we keep the build farm reporting accurate, but allow these other test modes to work. Andrew Bartlett
2009-09-08 04:48:40 +04:00
$(SELFTEST) $(SELFTEST_NOSLOW_OPTS) --socket-wrapper --testenv
r25707: Move selftest makefile code to separate file. (This used to be commit 9714db7814fd43d2a10485c041bbb761360c0826)
2007-10-22 03:05:31 +04:00
Use mkinclude (makes the Makefile a bit shorter), use double-colon for phony makefile targets. (This used to be commit ea7d389bcc284d2e9eeff79bce5c285269099ad3)
2008-02-09 18:05:55 +03:00
gdbtest:: gdbtest-all
r25707: Move selftest makefile code to separate file. (This used to be commit 9714db7814fd43d2a10485c041bbb761360c0826)
2007-10-22 03:05:31 +04:00
Use mkinclude (makes the Makefile a bit shorter), use double-colon for phony makefile targets. (This used to be commit ea7d389bcc284d2e9eeff79bce5c285269099ad3)
2008-02-09 18:05:55 +03:00
gdbtest-quick:: all
s4:selftest: rename SMBD_VALGRIND => SAMBA_VALGRIND metze
2009-03-20 18:11:14 +03:00
SAMBA_VALGRIND="xterm -n server -e $(selftestdir)/gdb_run $(LD_LIBPATH_OVERRIDE)" \
s4:selftest Fix 'make testenv-gdb' and 'make valgrindtest' etc The problem here was that tridge's changes to ensure that test results were always propogated didn't merge well with the addition of extra environment variables for the 'make valgrindtest' and similar environments. By splitting out the macro further, we keep the build farm reporting accurate, but allow these other test modes to work. Andrew Bartlett
2009-09-08 04:48:40 +04:00
$(SELFTEST) $(SELFTEST_QUICK_OPTS) --socket-wrapper $(TESTS) | $(FORMAT_TEST_OUTPUT) --immediate
r25707: Move selftest makefile code to separate file. (This used to be commit 9714db7814fd43d2a10485c041bbb761360c0826)
2007-10-22 03:05:31 +04:00
Use mkinclude (makes the Makefile a bit shorter), use double-colon for phony makefile targets. (This used to be commit ea7d389bcc284d2e9eeff79bce5c285269099ad3)
2008-02-09 18:05:55 +03:00
gdbtest-all:: everything
s4:selftest: rename SMBD_VALGRIND => SAMBA_VALGRIND metze
2009-03-20 18:11:14 +03:00
SAMBA_VALGRIND="xterm -n server -e $(selftestdir)/gdb_run $(LD_LIBPATH_OVERRIDE)" \
s4:selftest Fix 'make testenv-gdb' and 'make valgrindtest' etc The problem here was that tridge's changes to ensure that test results were always propogated didn't merge well with the addition of extra environment variables for the 'make valgrindtest' and similar environments. By splitting out the macro further, we keep the build farm reporting accurate, but allow these other test modes to work. Andrew Bartlett
2009-09-08 04:48:40 +04:00
$(SELFTEST) $(SELFTEST_NOSLOW_OPTS) --socket-wrapper $(TESTS) | $(FORMAT_TEST_OUTPUT) --immediate
r25707: Move selftest makefile code to separate file. (This used to be commit 9714db7814fd43d2a10485c041bbb761360c0826)
2007-10-22 03:05:31 +04:00
Use mkinclude (makes the Makefile a bit shorter), use double-colon for phony makefile targets. (This used to be commit ea7d389bcc284d2e9eeff79bce5c285269099ad3)
2008-02-09 18:05:55 +03:00
gdbtest-env:: everything
s4:selftest: rename SMBD_VALGRIND => SAMBA_VALGRIND metze
2009-03-20 18:11:14 +03:00
SAMBA_VALGRIND="xterm -n server -e $(selftestdir)/gdb_run $(LD_LIBPATH_OVERRIDE)" \
s4:selftest Fix 'make testenv-gdb' and 'make valgrindtest' etc The problem here was that tridge's changes to ensure that test results were always propogated didn't merge well with the addition of extra environment variables for the 'make valgrindtest' and similar environments. By splitting out the macro further, we keep the build farm reporting accurate, but allow these other test modes to work. Andrew Bartlett
2009-09-08 04:48:40 +04:00
$(SELFTEST) $(SELFTEST_NOSLOW_OPTS) --socket-wrapper --testenv
selftest: move gdb_* and valgrind_* scripts to selftest/ metze
2008-10-28 14:20:59 +03:00
Copy Permalink