Upayavira wrote:

Ralph Goers wrote:
Reinhard Poetz wrote:
yes, according to the mails above sometime in the future it will work.

                                  - o -

If somebody has time to fix gump.xml so that it builds at least
cocoon-core it would be a good idea. If not, we should simply ask the
Gump folks to remove the descriptor.

WDOT?

Everytime gump breaks I find myself wondering why anybody cares?  Can
someone educate me on what is better about gump then us running Continuum?

Firstly, I am no expert on Gump.

You can see Gump as more of a social thing - it not just related to our
own pretty little project - it builds _everything_. It checks out trunk
on a huge number of projects and builds them, thus ensuring that all
projects continue to work together in their trunk versions. So think of
it as one huge Continuum for _all_ Apache software and beyond, not just
for Cocoon.

As such, it will tell us if, for some reason, Cocoon wouldn't compile on
Kaffe, or if one of our dependencies changed its interface and that
broke our code.
How can that work? We specify the versions in the pom.xml. It will just keep rebuilding the same thing until the pom is changed. I guess I'm not sure how that even worked in the old system since we had all the jar files in our own repository.

Cocoon is a fantastic project for the Gump people, because we have soooo
many dependencies. If Cocoon builds, that means all of its dependencies,
and their dependencies, etc, all built too, which as I understand it
doesn't happen that often :-)
See the above comment. How does that happen if we still have the old version?

So, yes, it is a valuable resource, but on a broader scale than just our
own project.

Regards, Upayavira
Thanks. I sort of knew all that, but what I guess I'm missing is how it actually does what you are saying it does. With a maven2 build how do they "force" you to pick up the latest version (actually - I already know the answer to that since I have a Jira bug opened on Maven 2 to do just that. The answer is, you can't).

Ralph



Reply via email to