Unit Test Slowness [Re: 3.0 M3 test results]

From: Andrus Adamchik (andru..bjectstyle.org)
Date: Mon Jan 28 2008 - 17:00:45 EST

  • Next message: Andrus Adamchik: "Re: VOTE: 3.0M3 Release"

    I finally found time to run unit tests of the M3 tag on various DB's
    (evaluating release artifacts is next on my list).... I'll post the
    results in a little bit (so far all good), but I wanted to comment on
    the merge test case slowness.

    Because of repeated DB metadata access via DbLoader I noticed a bit of
    slowdown on MySQL. Then I started testing on Oracle, and this one is a
    bummer. My unit tests are already running for about 50 minutes with no
    end in sight... So maybe there is a way to cache the DbLoader calls
    for the purpose of unit testing? Would be nice if we could optimize
    that on trunk.

    Andrus

    On Jan 27, 2008, at 12:46 AM, Kevin Menard wrote:
    >
    > On 1/25/08 4:34 PM, "Kevin Menard" <kmenar..ervprise.com> wrote:
    >
    >> Windows XP Pro SP2 (Java 1.5)
    >> =============================
    >> HSQLDB: FAIL
    >>
    >> Only one test failed for HSQLDB for Windows and it was the following:
    >> testEncodeAsXML(org.apache.cayenne.map.ObjRelationshipTest)
    >>
    >> Without looking into it too much, it's probably a line ending issue.
    >
    > I think I'm going to address the line ending issues once and for all
    > by
    > setting the eol-style to native in Subversion. I'll just wait until
    > trunk
    > for this.
    >
    > There's also an issue with the JpaUnitClassLoader on Windows. So,
    > the JPA
    > itests all fail. While unfortunate, I don't think this is worth
    > holding the
    > release up. The other tests all pass, so I'm fairly confident the
    > core
    > framework is good to go.
    >
    > --
    > Kevin
    >
    >



    This archive was generated by hypermail 2.0.0 : Mon Jan 28 2008 - 17:01:18 EST