On Sat, 2005-07-30 at 23:36 +0200, joerg wrote: > -----BEGIN PGP SIGNED MESSAGE----- > Hash: SHA1 > > Hey there, > > I am back on the list after over one year. I have actively used maven1 > in various projects. I liked it but also suffered under some aspects. > > Now I had a look at m2: > > It seems that m2 will have the ability to categorize dependencies (test, > compile-time, run-time). This is awesome and was something I missed a > lot in m1.
Dependency scopes, yes. > It seems that m2 will support "recursive dependencies". This is also a > great idea. It might cause some trouble when e.g. a Project A depends on > B and C while B and C both depend on D but in different versions. We've got this covered with what we call dependency mediation where we can deal with conflicts. > I guess this problem is not new to you guyz and we all had and have such > problems with or without maven. Anyways I'd like to know your ideas how > to handle this issue in m2 (there should be a way for the m2 user to > manually decide which version to use - maybe simply by defining the > troublecausing dependency on the toplevel project in the desired > version). Anyways I think that "recursive dependencies" are in general a > good idea. There will be various strategies but the one employed now is the "nearest" where if you define a version of a dependency in your POM it will win. > It seems that sub-projects (in that "mulitproject" thingy) do not define > their dependency to its parent project by a filesystem reference but a > repository reference and the POM of the parent project that has to be > deployed there before. Again I have to say "great" - this will allow > you to check out only an single sub-project from a huge project > and build it directly from that point. Yes, that is possible but I think in practice you will probably still want to check out the whole tree, or a the whole group. Say all of Maven, or Geronimo. > It seems that you've decided to kick out jelly and build plugins > (usually) as native java classes. Concratulations! This is the best > thing you could do. Maybe m2 will be reasonable fast. At one point in the not too distant past it executed approximately 2x faster than Ant. I haven't checked the speed lately. > It seems however that m2 is a lot less documented than m1 (is that > possible:) ) and m2 is far beyond from being complete. Not really :-) We are drawing very close to being feature complete for 2.0. There are ~160 issues left but many of them are minor and you have to take into account the phenomenal rate at which they are being closed. That's Brett, John and many people who have started contributing to m2 in particular Kenney, Stephane and the two Vincents :-) > Now that's where we go. Since I seem to like your ideas I wonder > if I could get involved a little (do not expect too much cause I > am already involved in another open-source project - that is of cause > using maven). You can get as involved as you like :-) > Maybe you could help me out getting into m2 a little faster. > I do not have a clue what the actual state of m2 is (are you still > planning important architectural decicisions or is everyting clear > and just some code missing :)). Check out JIRA and you'll see the roadmap: http://jira.codehaus.org/secure/BrowseProject.jspa?id=10500 > You have some important links > not yet on the site (wiki?, ...). http://docs.codehaus.org/display/MAVEN/Home > I did not read deep into the history > of this dev-list so maybe you have a few important mails in some > mailfolder that you could just forward... > > BTW it seems that the POM inheritance in subprojects is not yet working > properly. I had to define things like folders (that src/main/java, ... > stuff) and specific settings (JDK 5.0 compliance, funny that it is still > - -source 1.5 and not 5.0) in every subproject. It's working well as far as we can tell, we have lots of unit tests and integration tests. > So now I will try to get a little into the source (understanding plexus > and all this kind of stuff) - might take me some weeks to become into, > however. Oh, you'll catch on I'm sure. Kenney figured the whole thing out in 12 minutes so that's the record you should be shooting for :-) > See you guyz out there! -- jvz. Jason van Zyl jason at maven.org http://maven.apache.org Three people can keep a secret provided two of them are dead. -- Unknown --------------------------------------------------------------------- To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
