On Wed, Aug 08 2012, Mark Walters <markwalters1009@gmail.com> wrote: >> test/dump-restore | 9 +++++++++ >> 1 file changed, 9 insertions(+) >> >> diff --git a/test/dump-restore b/test/dump-restore >> index 439e998..7979ebf 100755 >> --- a/test/dump-restore >> +++ b/test/dump-restore >> @@ -82,4 +82,13 @@ test_begin_subtest "dump outfile -- from:cworth" >> notmuch dump dump-outfile-dash-inbox.actual -- from:cworth >> test_expect_equal_file dump-cworth.expected dump-outfile-dash-inbox.actual >> >> +test_expect_success 'roundtripping random message-ids and tags' \ >> + 'test_subtest_known_broken && >> + ${TEST_DIRECTORY}/random-corpus --config-path=${NOTMUCH_CONFIG} && >> + notmuch dump > EXPECTED.$test_count && >> + notmuch tag -random-corpus tag:random-corpus && >> + notmuch restore < EXPECTED.$test_count 2>/dev/null && >> + notmuch dump > OUTPUT.$test_count && >> + test_cmp EXPECTED.$test_count OUTPUT.$test_count 1>/dev/null' > > Are the single quotes at the start and end of the main block meant to be > there? And with them deleted this seems to pass (but there is lots of > diff if the redirection is removed). I am not familiar with > test_expect_success/test_cmp so don't know what to expect. I don't understand what's going on here either. This seems like a strange way to run these tests, as a command string to test_expect_success. Why not just run them directly? I'm also worried about the test output blowing away the users terminal. I think that should be avoided, even if we expect failures to be rare. jamie.