Re: Hudson unit test runs

From: Andrus Adamchik (andru..bjectstyle.org)
Date: Fri Dec 11 2009 - 08:00:25 EST

  • Next message: Apache Hudson Server: "Hudson build is back to normal: Cayenne-trunk Β» HSQLDB,JDK 1.5 (latest) #93"

    I agree with the general statement (and indeed on many DB's we
    suppress certain tests, as the DB simply doesn't provide needed
    capability). But I think we need to classify this specific case
    differently, since Cayenne can still potentially generate the right
    single column SQL in this case.

    Andrus

    On Dec 11, 2009, at 7:52 AM, Andrey Razumovsky wrote:

    > We're testing Cayenne, not Derby.. Derby limitations is not our
    > fault..
    > ideally we should exclude that per DB, but I'd prefer that until we
    > have
    > such mechanism in our test framework, tests stiil run fine
    >
    > 2009/12/11 Andrus Adamchik <andru..bjectstyle.org>
    >
    >>
    >> On Dec 11, 2009, at 3:32 AM, Andrey Razumovsky wrote:
    >>
    >> OK, may I suggest we just slightly change the test, so that it
    >> works on
    >>> Derby as well. Current problem is that Derby does not allow more
    >>> than one
    >>> column in subquery result.
    >>> So if wwe replace
    >>> "SELECT a FROM Artist a WHERE EXISTS (SELECT DISTINCT p1 ...) "
    >>> with
    >>> "SELECT a FROM Artist a WHERE EXISTS (SELECT DISTINCT
    >>> p1.paintingTitle
    >>> ...)
    >>> "
    >>>
    >>> it will work with same sense
    >>>
    >>
    >> Then how do we know that the other syntax variant is broken?
    >>
    >> We clearly need an exclude list per DB...
    >>
    >> Andrus
    >>
    >>
    >
    >
    > --
    > Andrey



    This archive was generated by hypermail 2.0.0 : Fri Dec 11 2009 - 08:00:56 EST