Let's see what will be the result. It's interesting. But once again, -1 to change the default behavior.
Jean-Louis 2013/5/13 Romain Manni-Bucau <[email protected]> > well IMO we should go for 1 with a config to fallback to what we have now > to get a nice default > > i'll try to commit it to see how TCKs behave with such a config, then we'll > see i think > > *Romain Manni-Bucau* > *Twitter: @rmannibucau <https://twitter.com/rmannibucau>* > *Blog: **http://rmannibucau.wordpress.com/*< > http://rmannibucau.wordpress.com/> > *LinkedIn: **http://fr.linkedin.com/in/rmannibucau* > *Github: https://github.com/rmannibucau* > > > > 2013/5/13 David Blevins <[email protected]> > > > In a similar boat. I think #1 is a feature we need regardless, though I > > agree with JL that the default should be as it is now. > > > > I have a small hope that we could find a way to make managing shaded > > dependencies a bit easier. If we could find a cleaner way to do it than > we > > do now, might be more realistic to try. It's a bit of a project though. > > > > > > -David > > > > On May 13, 2013, at 1:31 PM, Romain Manni-Bucau <[email protected]> > > wrote: > > > > > i'm playing with 1 ATM, 2 would be the best but we'd need to maintain > > > it...impossible today > > > > > > a shade will not work without custom clever transformer (never a good > > > solution) > > > > > > so basically i'm for 1/ too if my tests show it works fine enough > > > > > > *Romain Manni-Bucau* > > > *Twitter: @rmannibucau <https://twitter.com/rmannibucau>* > > > *Blog: **http://rmannibucau.wordpress.com/*< > > http://rmannibucau.wordpress.com/> > > > *LinkedIn: **http://fr.linkedin.com/in/rmannibucau* > > > *Github: https://github.com/rmannibucau* > > > > > > > > > > > > 2013/5/13 Jean-Louis MONTEIRO <[email protected]> > > > > > >> 1/ just adding a property for an app with values like parent-first > > (default > > >> - today) and parent-last (new value). > > >> 2/ that usually the way other apps servers are used to do (IBM > > Websphere AS > > >> IIR) > > >> 3/Don't know > > >> 4/No > > >> > > >> The first one looks the more simple to implement (at first glance) and > > the > > >> one which produce less impacts and possible side effects for users. > > >> The second is something which can be studied furthermore and implement > > as a > > >> cherry on the cake. > > >> > > >> Then, if I had to choose, I would choose 1/ > > >> > > >> Jean-Louis > > >> > > >> > > >> > > >> > > >> 2013/5/13 Romain Manni-Bucau <[email protected]> > > >> > > >>> Hi > > >>> > > >>> as you probably know we changed quite often the classloading these > last > > >>> versions. > > >>> > > >>> We still have regularly issues with tomee provided libs (mainly > because > > >> of > > >>> cxf/amq/... transitive deps) > > >>> > > >>> I would like to go further and use an URLClassLoaderFirst for lib > part > > of > > >>> ear to allow apps to not rely on tomee libs first. > > >>> > > >>> The alternative is to shade/repackage all deps. > > >>> > > >>> PS: i want to avoid OSGi > > >>> > > >>> So basically here is the question: to solve conflict lib issue what > > your > > >>> preferred solution: > > >>> 1) URLClassLoaderFirst even for ear lib part - if this one please say > > if > > >>> you would like a property to switch back to old behavior > > >>> (openejb.ear.lib.classloader.first=false?) > > >>> 2) repackage all deps > > >>> 3) other > > >>> 4) OSGi/OSGi like > > >>> ? > > >>> > > >>> If nothing moves in the following days i'll go for 1 (then we can > still > > >>> discuss ;). > > >>> > > >>> *Romain Manni-Bucau* > > >>> *Twitter: @rmannibucau <https://twitter.com/rmannibucau>* > > >>> *Blog: **http://rmannibucau.wordpress.com/*< > > >>> http://rmannibucau.wordpress.com/> > > >>> *LinkedIn: **http://fr.linkedin.com/in/rmannibucau* > > >>> *Github: https://github.com/rmannibucau* > > >>> > > >> > > >> > > >> > > >> -- > > >> Jean-Louis > > >> > > > > > -- Jean-Louis
