Jason, I guess I am going to accept your generous offer even though as a fairly average Maven user I have not understood half of what you explained about the limitations and implications of your current prototype and a projected final solution.
Feel free to send me private messages whenever you are ready or just stay here on the list. I will follow your path and am hoping to at least provide some valuable feedback in addition to creating extra work on your side. -- Alexander Kriegisch > Am 01.04.2014 um 16:57 schrieb Jason van Zyl <[email protected]>: > I have a repository with a prototype if you want to take a look. I just need > to remove some crap and you're welcome to run with it if you want to do > something with it. The method employed works by mutating the model once it > has been read with a lifecycle participant. You can get pretty far using this > method but ultimately I think that the DefaultModelBuilder will need to be > changed. There are some limitations in having to deal with the fully > populated model where you can't do things like inject new properties and have > them be used because the ProjectBuilder has already interpolated them. We > probably could add some extensions points that would allow this post > processing to work but in the end I think it would be more efficient and a > better design to just change the ModelBuilder. > >> On Apr 1, 2014, at 4:42 AM, Alexander Kriegisch <[email protected]> >> wrote: >> I guess at least since 2008 Jason had the idea to implement Maven mixins, >> i.e. a kind of include mechanism in order for users to be able to >> encapsulate sets of dependencies (not just their version numbers as with >> import-scoped BoMs) and re-use them throughout modules or even across >> projects. This is something I (and probably many others) have been waiting >> for desperately because via inheritance you can only have one aspect >> covered, but not multiple ones. Comparing to single inheritance in OOP, in >> order to cleanly encapsulate secondary (even cross-cutting) concerns there >> are powerful concepts like AOP (I love to use AspectJ for that matter) or >> simpler ones like mixins. I think mixins would suffice for Maven and >> massively support DRY (don't repeat yourself), thus making it much easier to >> keep Maven configurations cleaner and more maintainable. >> >> Having said that, I know it is always easier to ask for something than to >> just implement it by myself (which I am uncapable of because I have never >> touched or even seen the Maven code base). So, humbly and with all due >> respect to you as a newbie on this list, I am asking if mixins are still on >> the development agenda and if the feature has a rather low or high priority. --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
