Re: Upgrading ObjectStyle.org Confluence

From: Boris Kraft (boris.kraf..binary.com)
Date: Fri Nov 11 2005 - 05:54:41 EST

  • Next message: Boris Kraft: "Re: Upgrading ObjectStyle.org Confluence"

    On 11.11.2005, at 10:26, Andrus Adamchik wrote:

    > Boris,
    >
    > Thanks for the info.

    always welcome

    > I am already feeling dizzy from looking at all the choices,

    its a mess out there

    > and still I don't see one that will just work for us without too much
    > pain. The fact that Magnolia is in Java (and open source) is a
    > definite plus.

    Exactly. It lets you do all the things you want

    > I don't mind that it is not WebObjects (Cayenne is not WebObjects
    > either). (JSP scares me though, but that's beyond the point :-)).

    Right, but the plus side is that you can easily hack *whatever* you
    want in Java quickly.

    > A question... I am not familiar with JSR-170, but I don't see any
    > mention of a database backend anywhere. How is the data stored? Even
    > if this is something abstracted by the JSR, it has to reside somewhere
    > and be searchable.

    The JSR-170 API abstracts from the actual storage, which is one of the
    huge benefits. For all our websites and applications based on Magnolia,
    we can simply switch the backend repository implementation by
    installing a new version (of Magnolia), configuring to use the
    repository we like, and then activate the content from one Magnolia to
    the next.

    Since the JSR-170 market is very much in movement, thats pretty cool.
    As soon as there is a better implementation - just switch.

    We started with a filesystem based implementation (XML), then CQFS (a
    proprietary database), now its berkley DB ... all beneath Jackrabbit,
    an Apache project for the API implementation. There are also commercial
    repositories (Day, Exo) and anothzer open source one: Jeceira.

    So there will be more and more chices that offer additional benefits -
    full text indexing or webDAV access are good examples.

    For the search as such, the API lets you choose between JCR-SQL
    (basically SQL) and XPath, but AFAIR thats repository dependend (I
    thing the last spec was that XPath is mandatroy and SQL optional, but
    it changed so often I can't remember) Anyways Jackrabbit suports both.

    Finally, just for completeness sake - its no good idea to access the
    storage system directly. Use JSR-170, thats what its for.

    HTH
    Boris



    This archive was generated by hypermail 2.0.0 : Fri Nov 11 2005 - 05:54:54 EST