That code hasn't changed basically since version 1 ... It has always
been a little bit sketchy. The problem is that you can reference
models all over the place, so coming up with a consistent and cross-
machine way of resolving the path to them is annoyingly tricky. The
current implementation "guesses" when it should try to build a
relative path and when it should just fall back and choose an
absolute path. I think the rule is something like whether or not
there is any part of the two paths that they have in common. So
Chuck would probably be right that if you are crossing volumes, it
would have a hard time.
This has always bugged me. My current thoughts are that it should
just name the models and let model resolution find them itself. The
problem with this is the same issue we ran into with Entity Modeler
where people don't necessarily have uniquely named models in
development mode (this generally seems to be when people don't use
prototypes and instead opt for their own method). In that case, you
would have to also have some mechanism for defining which specific
models you mean. For Entity Modeler we actually use the same editor
as eogen files so you can pick models exactly BECAUSE it supports
path references for disambiguation. If eogen files removed that
feature, then there would have to be both a new method for
creating .eomodelgroup files as well as a new method to
disambiguate .eogen files for people who have non-unique names.
My point here is that this problem is a big pain in the butt :), and
I don't have a particularly good answer for it at the moment.
ms
On Jan 11, 2007, at 11:08 PM, Chuck Hill wrote:
>
> On Jan 11, 2007, at 5:55 PM, Lachlan Deck wrote:
>
>>
>> On 12/01/2007, at 12:48 PM, Chuck Hill wrote:
>>
>>> I have never gotten it to generate an absolute path. It always
>>> makes them relative for me. What are you doing?
>>
>> Have you looked at the *.eogen file in a text editor afterwards?
>>
>> All I'm doing is clicking on "Add..." and selecting a referenced
>> model.
>
> I just tried it and this is what it added:
>
> -refmodel ../../../../../GVCSiteMaker/Code/GVCSMCore/
> GVCSiteMaker.eomodeld
>
> (this is across projects, clients, etc, so rather an extended
> relative path)
>
> Are both of your paths on the same volume? I am on 3667, maybe you
> are on a new, broken version?
>
> Chuck
>
>
> --
>
> Practical WebObjects - for developers who want to increase their
> overall knowledge of WebObjects or who are trying to solve specific
> problems.
> http://www.global-village.net/products/practical_webobjects
>
>
>
>
>
This archive was generated by hypermail 2.0.0 : Fri Jan 12 2007 - 09:29:43 EST