Jean-Baptiste Quenot skrev:
And FWIW I added an interesting topic for the Hackaton:
See http://wiki.apache.org/cocoon/GT2006Hackaton
-->
[Jean-Baptiste Quenot] Try to replace Maven 2 with Ant and Ivy
You are of course free to experiment with whatever you feel like. But
IMO it would be to attack the problem from the wrong direction. The main
problem with Cocoon from a build POV, is that it is so huge and has so
many dependencies. This is the problem that we should attack.
As we have discussed many times we should split up the core, so that it
only contains the absolute minimum and move everything else to optional
blocks. This would considerably reduce the number of dependencies for
users, and as long as the number of dependencies is reasonable Maven is
OK. We could also move the majority of the blocks to a "legacy" area and
remove them from the main build. For most Cocoon developers there is no
use at all to build all the +100 modules, it would be enough with the
core and a few other blocks. For the rest it is enough that specially
interested people and the Continuum integration build builds them, so
that we will see if we happen to break anything there.
So IMO it would be better to invest our efforts in the *root* cause of
our build problems, i.e. decreasing the complexity of the dependencies
rather than trying to hack around them by spending even more effort in
using new build systems and forcing everyone to learn still another
build technology.
/Daniel
