https://bz.apache.org/SpamAssassin/show_bug.cgi?id=8033
--- Comment #5 from Sidney Markowitz <[email protected]> --- (In reply to Loren Wilton from comment #4) > What's the time to run all tests? On the same VM that measured the 2 minutes difference, the default tests ran in 28 minutes. If I set the options to not run the 31 tests labeled as "long" instead of the default, the remaining 183 tests run in 14.5 minutes, for an average of under 5 seconds per "short" test. That puts bayessql.t in the range of the long tests, even using /dev/shm at 23 seconds compared to the average of under 5 seconds for all the short tests. So one change should be to label it as long. That would make it still longer than the average time for the "long" tests, but still, adding 2 minutes to a total test suite of half an hour doesn't seem like too much to me. Thanks, I think that puts it in perspective. If this next round of tests I submitted comes back showing that writing to disk instead of /dev/shm makes it work on those test machines, I'm inclined to make that change and label it as one of the long tests. -- You are receiving this mail because: You are the assignee for the bug.
