Hi Leo, On 4/17/05, Leo Simons <[EMAIL PROTECTED]> wrote: > The more and more I look at this, the more I'm disliking how all this is set > up, esp. as its not very consistent across different projects, and we don't > have very clear guidelines on how people should be doing this (other than > "copy existing practices"). It's a little messy. > > I'm tempted to do a radical remodelling of our metadata structure to remove > this kind of ambiguity, even going as far as having conventions like > project-name-is-file-name be gently enforced.
This was exactly the problem that was encountered in attempting to match gump IDs to Maven IDs - firstly that there was inconsistent naming, but also because they used a different scheme. I've detailed this, and possible solutions in either Maven or Gump here: http://wiki.apache.org/gump/MavenId What do you think of using the group ID and artifact ID ideas in gump? So: - a group is a collection of projects, and a project builds one or more artifacts. - group probably equates to a repository, currently - project is a build, but has no ID of its own, just a path relative to the repository - artifact equates to the jar id in gump now - all internal references are by groupID + artifactID Note also that groups are hierachical now, so while jakarta-commons may be a group, so is jakarta-commons-jelly (as long as a group can contain a group, this will work from gump perspective also). WDYT? Cheers, Brett --------------------------------------------------------------------- To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
