On 2/13/2013 7:10 AM, Peter Klügl wrote: > On 11.02.2013 22:20, Richard Eckart de Castilho wrote: >> Am 11.02.2013 um 22:01 schrieb Marshall Schor <[email protected]>: >> >>> Thanks for taking this on :-) >>> >>> I thought these artifacts were Eclipse packages - Don't these (usually) have >>> source? >> Well, they must have sources unless somebody out there is writing byte-code >> by hand ;) The problem is that the "eclipse:to-maven" goal doesn't convert >> the sources. There appears to be an interesting project called >> "maven-osgi-repo" [1] which can claims to support the conversion of Eclipse >> source bundles as well. It doesn't say anything about JavaDoc, but that >> should be doable once the source is there. >> >> Looking into my Eclipse "plugins" folder, there are some bundles that have a >> companion "source" bundle, and others that do not. It may be possible to set >> up an Eclipse target definition that includes all the bundles required by >> TextMarker including their source bundles. I don't know if source bundles are >> available for all of these bundles, though. >> >> The next sensible steps would be to get a target platform set up, try the >> maven-osgi-repo, and hope all the sources are available and all the bundles >> are convertible. If that works out, something would need to be done to build >> the JavaDoc artifacts. Ant may come in handy. >> >> Anybody is welcome to step in and try things. Otherwise, I'll keep this on >> the back burner, getting back to it as time permits and of course posting any >> news. >> >> Cheers, > > Thanks Richard for digging into this. > > maven-osgi-repo sounds great, but is it an applicable solution for us? Would > ASF servers host it? > > I always get back to the idea of using tycho, which would solve all related > problems. However, it will create new ones since it requires some changes that > are not in line with how the other projects are structured, e.g., the artifact > id needs to correspond to the bundle id/name. > > If I understood the answer to Marshalls question correctly, then the artifacts > hosted by Richard are not a critical blocker, right?
I think it's not a blocker, but it's unstable, a bit - because it's more likely that Richard's hosting will (at some point in time) go away, and then we can't build... It would be great to get these into Maven Central... -Marshall > > Peter > > >> -- Richard >> >> [1] http://code.google.com/p/maven-osgi-repo/ >> > >
