Re: EOGenerator Replacement

From: Mike Kienenberger (mkienen..mail.com)
Date: Mon Mar 12 2007 - 15:48:28 EDT

  • Next message: Chuck Hill: "Re: EOGenerator Replacement"

    Mike, I completely agree with you. This is what I meant about
    usability issues and EOModels being second-class citizens. You'd be
    able to import a Cayenne model (you can do this already with only a
    little bit of information loss), but then you'd have to maintain it
    from a "Cayenne" mindset. The "save as" action would then translate
    the Cayenne stuff back into EOModel stuff (reversing the import
    conversions). So you'd be "programming" your model as if it were a
    Cayenne model, but storing it on disk as an EOModel. I don't think
    it'd really affect the Cayenne modeler code much -- the code for doing
    the import/export conversions would be localized to those actions.
    However, the end user would have to be both Cayenne- and an
    EOF-model-aware.

    I don't think anyone is going to take this on because of the limited
    utility of it. But I think once every part of a EOF model has been
    represented in the Cayenne model, someone will take the 30 minutes
    required to output 95% of the data in EOF model format. It'll
    probably be someone who's experienced at using Cayenne that has to
    maintain an old EOF model. Like me :-)

    Here's hoping that by that point Entity Modeler is so advanced that
    there's no need for it :-)

    On 3/12/07, Mike Schrag <mschra..dimension.com> wrote:
    > I think the things that will get you are concepts that are KIND OF
    > close, but not EXACTLY the same. EOF obviously has very specific
    > (and some very strange) semantics for things that you would have to
    > match in the Cayenne model. I suspect this would make the Cayenne
    > code progressively crappier as you add all these weird little special
    > cases. Two that come to mind right now are EOQualifiers and
    > Cayenne's query system -- close but not an exact match (some of the
    > keywords aren't the same) and things like the stored procedure stuff
    > in EOM format for fetching/inserting/etc (combined the weird ordering
    > semantics of the stored proc attributes). It's deceptively close,
    > but Cayenne has a nice, fresh view of things -- it would be like
    > tracking mud through a clean kitchen :) But hey, if someone wants to
    > take it on, more power ...
    >
    > More on topic, though, I'm all for bringing WOGen up to speed. It's
    > been a checklist item on my perpetually growing checklist, but "real
    > work" has gotten in the way the past couple months.
    >
    > ms
    >
    > On Mar 12, 2007, at 3:04 PM, Mike Kienenberger wrote:
    >
    > > On 3/12/07, Mike Kienenberger <mkienen..mail.com> wrote:
    > >> The only missing functionality is already on the TODO list
    > >> for the standard Cayenne modeler (prototypes, user properties, uml
    > >> diagramming, certain kinds of inheritance).
    > >
    > > And derived types. Ok, the list keeps getting longer :-) Actually,
    > > if it's just a matter of equivalency, prototypes and uml stuff isn't
    > > required.
    > >
    > > As I said before, I'm willing to be a resource for porting Cayenne's
    > > cgen stuff as an EOGenerator replacement. I might even have my
    > > scripts still around somewhere that I used to convert my EOGenerator
    > > templates into velocity templates. I suspect I wrote them as
    > > MiscMerge templates. :-)
    >
    >
    >



    This archive was generated by hypermail 2.0.0 : Mon Mar 12 2007 - 15:48:42 EDT