On Nov 6, 2008, at 3:44 PM, Thorsten Scherler wrote:
Hi everybody,
like Oleg pointed out a couple of time, droids is starting to provide
different components and maven can help to manage complex multi-module
projects.
ATM our ant scripts are kind of broken since we moved lots of things
around. The question for me now is whether it makes sense to fix it or
just drop the ant scripts and go 100% maven.
I am not a huge maven fan... and find myself lost when things don't go
right... however, given the multi-module aspect of droids, i think
maven is the better choice.
If we use maven as the base build system, we can still use ant for
tasks that are not supported in maven. See:
http://maven.apache.org/guides/mini/guide-using-ant.html
One possible problem is that ATM we use quite a lot of build
properties
files I am not sure how we can incorporate them into maven. Further I
personally can not be a big help in the move to maven, since I need to
get into maven again.
I don't see any big issue translating the build properties into
maven. In fact, I see a big win in separating the "build" properties
from the "config" properties. For example:
javadoc.link.java=http://java.sun.com/j2se/1.4.2/docs/api/
vs
droids.delay.request=500
The only part I am not sure about is forrest integration -- I have not
done it, and a quick internet search does not immediately answer any
questions. Perhaps we may need to use the maven ant integration to
call forrest task...
IMO maintaining two different build system offers few benefits.
WDYT?
I think we should try to build a functioning pom.xml and then evaluate.
Oleg, if you are comfortable with maven, can you give that a crack?
If not -- i will try to throw something together
ryan
