Jason van Zyl wrote:
>
> I think that Gump would be used to feed the repository with the latest JARs,
> but I don't think most people would be interested in layout of the projects.
> The just want the docs and build things with it.
No, the repository should contain stable and tested versions. IMHO.
One thing you may not realize is that I have been using Gump for my daily
development since the beginning of this year. Mostly, I've been working on
a project named xml-axis. It depends on xerces, ant, servletAPI and
others. I find those jars based on the project definitions.
> Yes, I understand :-) I would like to use the dependency information gather
> by gump and allow JJAR to use this, but I think the information would have
> to change a bit. I will know more when I get the first example working. We
> also need a way to move these project descriptors into the project. If we
> potentially get hundreds of projects this will become a loathsome task, and
> I think it's the projects responsibility. But I definitely want to use the
> idea of sharing project descriptors for mechanisms like gump and an
> automated build system.
That's why it is <project href="...">. URL's need not all be relative...
;-)
I keep trying to goad you into building a tdk profile, and try it with Gump
before you go off reinventing, but so far it hasn't worked. I even have
been trying to address all the issues that you have brought up...
> Right now the project controls what it wants to use in a project level
> build.properties file. I don't see the need for a workspace for a project
> build.
You will. ;-)
Now that turbine and avalon have exploded, I was hoping that more people
would recognize that they rarely work on a single project, but rather on
several projects in combination. I bet that is true of everyone who
contributes to jakarta-commons.
- Sam Ruby
