https://bz.apache.org/SpamAssassin/show_bug.cgi?id=8033
Sidney Markowitz <sid...@sidney.com> changed:
What |Removed |Added
----------------------------------------------------------------------------
Resolution|--- |FIXED
Status|NEW |RESOLVED
--- Comment #6 from Sidney Markowitz <sid...@sidney.com> ---
The test results came back, with seven runs in the set this time showing the
failures. On all of them, I logged the creation of the temp directory, the
calls to sa-learn that used it, and tested for existence of the directory at
various points in sa-learn. The directory disappeared between 2 and 6 seconds
after creation, all after variously between one to five calls to sa-learn
completed successfully. On all of those test runs, a copy of bayessql.t that
was identical except for the temp directory being on disk instead of /dev/shm
passed all tests.
I still don't know what those machines have in common that make them part of a
set and have that failure, but it doesn't seem worth figuring out just to speed
up this test on machines that have a /dev/shm.
trunk % svn ci -m "bug 8033 - Remove use of /dev/shm to speed up test because
that causes test failure on some machines. Label test as long running"
t/bayessql.t
Sending t/bayessql.t
Transmitting file data .done
Committing transaction...
Committed revision 1903870.
--
You are receiving this mail because:
You are the assignee for the bug.
