I have seen all the posts on this thread and it seems that a move to ant
is in
favor of people that replied.
Despite of some shortcomings of maven-1 (documentation/plugins) it's still
a powerful tool especially the dependency check. We use maven in our company
and an advantage was that it was simple to add a new project by defining the
dependencies (project.xml) and use the default tasks (war) for a build.
The general trend for apache projects builds is to move to maven-2.
The response so far was positive and lot of maven -1 issues are
resolved with maven-2. If we change the build tool it should be
maven-2 rather than ant.
In my opinion the major issues of the build process are not the tools
but the
complexity of the project. To many dependencies to internal components.
I don't think ant would be a good tool to manage that. With the current
complexity ant's build.xml would become un-manageable.
There was an email thread to simplify the structure and make the build
process
more manageable. We should continue on that.
My suggestion is the following:
* Simplify the complexity of the Jetspeed build.
* Decide if we stay with maven-1 or go with maven-2
Roger
Randy Watler wrote:
All,
We now have a marginal Maven2 build that is capable of building J2 and
installing on Tomcat. While it has been fun reinventing the wheel for
the Nth time, it is time to get serious about the J2 build. Here are the
options:
1. Continue on with Maven1/J2 plugin.
2. Step up and complete the Maven2 build and create an archetype to
replace the genapp capabilities.
3. Ditch maven and go with Ant.
We need to vote on this before I or anyone else puts more sunk time into
the build. Here are some of the issues:
1. Ant is simple and everyone understands it.
2. Maven1 and the plugin are not stable and are generally complex.
3. Maven2 has simplified things in some ways, but made them more complex
in other ways with the pom.xml inheritance and transitive dependencies.
4. Ant build.xml files can become unmanageable.
5. Maven2 may not be sufficiently mature for our use; we have
encountered several bugs and have used some ugly workarounds for even
our simple build cases handled to date.
6. J2 users have not been exposed to maven, and it can become a
liability quickly since they expect Ant like builds.
7. All IDEs, including Eclipse, can natively build Ant based projects.
8. When the BSR or other repos are down, the Maven offline builds are
hopeless.
9. The training/learning curve with maven is hurting acceptance of the
J2 portal solution.
10. The repository in Maven2 will become even more difficult to manage
with the transitive dependencies: in the end, we will be forced to
manage our own repository and all of the J2 users will need to do the
same.
I am sure there are more... this is not exactly a new topic for any of
us. We are just at the point where we need to make a final decision that
can stand the test of time... J2 needs our cycles, not the build
environment. I am willing to put more time into the build no matter
which way we choose to go... but not unless there is a consensus on the
matter.
Randy
---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
