I think Geir is really onto something here. I spent a couple of years working on a project whose goal was to facilitate software reuse across the scientific computing community, which often uses a development model similar to open source. Our initial approach was to create a monolithic repository for all the software files and this led to a number of problems, everything from issues with intellectual property rights to debates on look and feel (sound familiar?). The more successful approach was to create a customizable data model for the software metadata and to harvest the entire collection of metadata into a centralized repository that contained pointers to the files hosted elsewhere. Here's the project url with papers 'n stuff in case anyone is interested in reading further http://icl.cs.utk.edu/rib/
Best wishes, Paul On 5/2/06, Geir Magnusson Jr <[EMAIL PROTECTED]> wrote:
Aaron Mulder wrote: > I have to disagree with putting up an ASF option as the default. > > Let's say there are 50 plugins produced by Apache and 70 by outsiders. > > We have a choice to make the default a repository containing 50 > entries, or a repository containing 120 entries. What makes sense? Here's a alternative idea... How about hosting the directory/metadata of plugins at the ASF (or even cooler, do something mirrored to avoid the ire of infra when Geronimo is ubiquitous) and just have URLs to the plugin locations...? Then that drives all plug-in authors to come and "register" them here - just send a message to the mail list to have it included... Then it doesn't matter - you can list plugins under all licenses (including proprietary) - and they are hosted where they are hosted, if you know what I mean. No worries about Apache hosting things that aren't from the ASF, etc. geir
