On 10/08/2008, at 12:31 AM, Andrus Adamchik wrote:
This suggestion is the most damaging, as it would force us to break
out of our namespace. And everything else is chained to that. So
while I am not completely against a reorg, I'd say we should do it
as the last resort measure. Would it be simpler to use the other
plugin?
For sure, we should only do it if it helps us in some way, since any
change will break artifacts other people may be relying on. On the
other hand, maven is complex to navigate and I'd hate to think that an
unusual naming scheme discourages people from contributing to Cayenne
because they can't get the build environment working easily.
I'm not sure what 'other plugin' you refer to here.
Even after my Eclipse plugin fixes, I've still got the following issues:
1. "mvn cayenne:modeler" does not work (CAY-1058)
2. "mvn clean install" is required, not just once to bootstrap but
every time I want to get recent changes into an assembly. So,
packaging up the assembly does not detect changes to the source. In
order to capture new jars they not only have to be rebuilt but also
installed into .m2.
3. "mvn clean install" is not even enough. "mvn -P mac clean install"
is required for me to get the OSX specific bits compiled right.
4. Still, even with all the above, odd things happen which are no
doubt due to my cursory knowledge of maven's build process. But it
means that I spent an inordinate amount of time trying to get the
build system working after by upgrade to the latest Eclipse: probably
about 6 hours in the last few days. Now, if only I'd spent those 6
hours working with Marcin on inheritance or improving the Cayenne
documentation...
So, in short. If the reorganisation makes it simpler (for example, so
bootstrapping is not required) or more reliable (so that old code
doesn't get included), then we should do it. If I'm the only one
having troubles, then I'll just have to put in more work to understand
maven better and work around its quirks.
I know in the past, the committers here have been lukewarm about
maven, but no one wanted to put in the work to return to ant (fair
enough!). Is it worth considering ant+ivy as a new approach which
might deliver the best of both worlds? If so, I'm happy to work on
this over a period of time. But I wouldn't want to start if that
wasn't a communal long term goal.
Ari
-------------------------->
ish
http://www.ish.com.au
Level 1, 30 Wilson Street Newtown 2042 Australia
phone +61 2 9550 5001 fax +61 2 9550 4001
GPG fingerprint CBFB 84B4 738D 4E87 5E5C 5EFA EF6A 7D2E 3E49 102A
