On Thu, Apr 17, 2008 at 8:59 PM, Archie Cobbs <[EMAIL PROTECTED]> wrote:
> On Thu, Apr 17, 2008 at 1:10 PM, Xavier Hanin <[EMAIL PROTECTED]> > wrote: > > > It depends how you consider the roundup repository. I think its intent > is > > to > > be a kind of meta repository: it contains metadata helping to build an > > actual repository. Maybe what's confusing is that archie suggest to use > it > > as a repository directly. IMO this is required only until we can get > > enough > > support to have a good hosting for the actual repository. Then both the > > meta > > repository and the actual repository would make sense: the meta > repository > > is helpful for people who want to create their own repository, and > easier > > to > > update than a repository made by hand. But it isn't used directly from > > Ivy. > > The actual repository is the only thing that is used by mere users. > > > > There are actually three possibilities... > > 1. Builder Repository: repository contains ivy.xml and builder.xml > files only, and the user is required to use the builder resolver, > whereby > the actual artifacts are downloaded and extracted as necessary at > resolution > time. > 2. Builder Repository + Artifact repository: same as above, but you > configure your resolver with the "resourceURL" attribute (which > overrides > the normal resource URLs), which points to the Artifact repository > containing all of the ZIP/TGZ files that you need to download. The > Artifact > repository could be local or community/online. This means we don't have > to > rely on each/every project's server being reliable. > 3. Build-your-own-Repository: we would add a new ant task in the Ivy > RoundUp build.xml that actually executes all the builder.xml > download/extract instructions for every module in the repository to get > all > the artifacts and then combines those artifacts with the ivy.xml files > to > create a normal Ivy repository. > > Correlating with Xavier's comments: #3 == "meta repository" whereas #1 = > "repository directly". > > The important concept here is *separation of the meta-data from the data*. > > Also important is *not forcing the requirement for a high capacity, high > bandwidth public server*.... I'll repeat my claim that this is important > based on the evidence that nobody has yet volunteered to replace Ivyrep. > > If someone is willing to host a normal repository, then options #2 or #3 > make sense. If not, then #1 makes sense. But it is flexible enough to > work > in any/all of these ways. I think we should start with #1 and then we can > move to #2 or #3 if and when someone volunteers the additional servers and > disk space. +1. Starting with #1 is the best option right now, then we'll later move to either #2 or #3 (I prefer #3, but it's only my opinion). Xavier > > > -Archie > > -- > Archie L. Cobbs > -- Xavier Hanin - Independent Java Consultant http://xhab.blogspot.com/ http://ant.apache.org/ivy/ http://www.xoocode.org/
