On Tuesday 23 November 2004 04:46, Brett Porter wrote: Hi Brett. You suggested I dive in so I'm going to <grin>
> > I think this situation is best covered by publishing to different > repositories for each, and then only utilising the remote > repositories appropriate when building. IF you happen to specify > more than one remote repository, you get the newest which should > be correct. Not sure I follow here. Most of the time developers will work against the artifacts in the QA repository. In some instances they will want to work against QA artifacts except for those artifacts associated with a specific group (or possibly a specific artifact). Using a simple list of repositories with first one that matches wins won't work(unless there is a way to tag an artifact with a user specified identifier) > > There are two things we are aiming highly to do: > - reduce clutter in the POM (by making sure it is both concise > and has sensible defaults) > - ensure dependency specifications match or are closely tied to > the project definition so that IDs can be matched. > > This is why model changes in general, but dependencies in > particular, get heavy scrutiny before inclusion. Absolutely concur. I did some initial work on a declaritive build system about a year ago that featured something very similar to the POM. The problem was always making sure the POM did not become as or more complex than the Ant files it replaced. Also, per our discussion previously I am beginning the process of determining Cobalt's attitude to more direct involvement on my part. We use open source in our products and the organization has expressd a willingness to contribute in the past. Initial indications are good. -steve --------------------------------------------------------------------- To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
