Re: Modeler thoughts ...

From: Michael Gentry (mgentr..asslight.net)
Date: Fri Aug 21 2009 - 09:13:41 EDT

  • Next message: Andrus Adamchik: "Re: Modeler thoughts ..."

    On Fri, Aug 21, 2009 at 3:00 AM, Andrus Adamchik<andru..bjectstyle.org> wrote:
    > Hi Michael,
    >
    > thanks for this list. Always good to see how the end users view things vs.
    > your own assumptions about how it should work.
    >
    > On Aug 20, 2009, at 10:56 PM, Michael Gentry wrote:
    >>
    >> [1] E/H - Found creating a new model confusing.
    >
    > Could be a valid concern especially for people who have to unlearn stuff.
    > The existing tutorial should help, but of course it is not built in the
    > Modeler. I recommend the new users to spend half an hour on the tutorial.
    > This usually answers all these questions.

    I think if there was a wizard style create project, they'd be happier.
     If nothing else, create the DataDomain, DataMap, DataNode for them
    (ask for the values and explain what each is used for). Maybe ask a
    few things like PK generation strategy, etc, too.

    >> [2] E - Wanted the OS X concept of a wrapper/bundle for the model
    >> files.  Essentially, he wanted to be able to name his model file
    >> (something other than cayenne.xml), but as we talked more he wanted a
    >> bundle that he can name ("MyProject" -> MyProject.cayenne directory)
    >> and then the cayenne.xml, etc files would be under there.
    >
    > Somewhat related to that, there's a plan to allow multiple mappings from
    > unrelated projects to get merged in runtime, so better modularity can be
    > achieved. As for placing mappings in a folder, we've had this discussion
    > before. I don't see much point to it.

    Yeah, I know we've talked about it before, but I'm not the only one
    that likes the idea. :-)

    >> [3] H - One guy used Cayenne quite a bit, but eventually ditched the
    >> modeler and edited the XML by hand.  "It's not intuitive.  I was
    >> confused.  I remember having so many problems setting up the correct
    >> path so when files were generated they had the right path and package
    >> names."  I'm not sure which version of the modeler he was using, but
    >> probably 2.x.
    >
    > This was a long standing bug. It is fixed in 3.0M6. But if his only problem
    > was class generation, I don't understand why he ditched the Modeler? I use
    > the Modeler for all work *except* class generation, for which I'd use
    > Ant/Maven.

    I got that information second-hand and can try to glean more details
    later as to why he ditched the modeler. If some people are just
    naturally wired to want to edit XML by hand, we can't do much about
    that. However, if there were deeper issues, it'll be good to know
    them. I'll try to find out more and report back.

    >> [4] M/E/H - Allow multiple projects to be open at the same time.  The
    >> EOF guy considered this a bug.
    >
    > +1.
    >
    >> [5] H - More useful validation.  "When it is validating the mapping
    >> file, I dont find its output useful at all.  Since the mapping works
    >> despite the warnings raised in the modeler, I question the value of
    >> those warnings."  I think he was probably using 2.x then.
    >
    > Not sure what he means. Some examples would be helpful. Warnings are very
    > useful to me at least. All of them are important.

    I'll ask him for examples.

    >> [8] E - "I can't figure out how to flatten a relationship."  I had to
    >> fess up and tell him that I didn't know, either.
    >
    > Cayenne is different from EOF in having separate DB and object layers of
    > metadata, so naturally the process will be different. Use inspector button
    > on the ObjRelationship panel to set a db path (current trunk also includes
    > support for attribute flattenning via a similar UI).

    I can talk to him again, but he even went to the online docs and tried
    to figure it out and still wasn't able to do the mapping.

    >> [10] M - If you select an ObjEntity and look at the attributes, then
    >> resize a column (say, Java Type) so you can read everything in the
    >> column, then select something else, then come back to that same
    >> ObjEntity, the column doesn't retain the size you had set.  I suspect
    >> this applies to more than just the attributes tab (too lazy to go try
    >> the others right now).
    >
    > +1. There's a Jira being implemented right now that would memorize the
    > column width for all tables.
    >
    > Andrus

    Thanks!

    mrg



    This archive was generated by hypermail 2.0.0 : Fri Aug 21 2009 - 09:14:17 EDT