Jorg Heymans wrote: > Joerg Heinicke wrote: > >>>> Eeeeh, I hope not! I'd worry about the health of the overall >>>> codebase if >>>> no-one else is building allblocks! Is this why continuum is not failing >>>> when it should be? > > ... > >> >> I can only agree with Andrew. The goal has to be that all blocks are >> working and continuum runs all blocks. That's orthogonal to >> independent release cycles and community interest. As long it the >> blocks are in Cocoon's codebase we have to care for them. > > > FWIW continuum hasn't been running for a good few weeks now. I disabled > it around the time of Reinhard's POM refactorings and never got around > reinstalling it. > > The suggestion at the time was to let CI do a top-level build with > -Pallblocks , instead of having it parse the pom and build each block > separately. I believe Giacomo had a setup like this for his project. >
but I don't think continuum would have noticed this particular issue, would it? see [1]: it is fundamentally about not finding a 1.0-RC (not snapshot!) version of commons-jci that is however _only_ published on apache's snapshot(!) repository so all the dependencies are put down correctly, and the classpaths are build up as should. And those will actually point to a file on your local maven repo IIF by some luck you happen to have that 1.0-RC jar down there. which is the case for those having built the mildly unrelated 'cocoon-core' project in the time-window when: 1/ cocoon-core still had a dependency to that same version of the jci-jar 2/ cocoon-core already had the (still lingering) hack-patch to seek for non-snapshot stuff also on the snapshot repo) This is not the case for people that never built cocoon (or any other project that required jci-1.0-rc). In fact it would most likely be the case for continuum who's building the project ever so regularly... [1] http://mail-archives.apache.org/mod_mbox/cocoon-dev/200704.mbox/[EMAIL PROTECTED] -marc= (convinced we should at least patch something to avoid this)
