On Mon, 2003-11-17 at 10:20, Vincent Massol wrote: > Guys, sorry to ask again but I have trouble understanding what's in > those branches.
In rc1-refactor I attempted to fix everything which included making Maven a bean, removing Jelly/Ant from all code except the plugin manager, classloader separation, removal of betwixt with xpp3, proper inheritance across the board and a reimplementation of the plugin cache manager. I finished and everything bootstrapped but there is a show stopping problem pre/post goals as a result of me revamping the way werkz handled goals. The changes in Werkz are necessary in order to isolate the classloader a particular plugin uses. I still think all of these things are required but currently I don't really care that much about my branch as all I really want is something we can call 1.0. Ultimately all of the code is going to be replaced in a piecewise fashion with well tested components. > What is the status with properties inheritance? It works in my branch because I removed Jelly from equation which makes things easy. These bits I might merge into Brett's branch because you need to remove Jelly/Ant from the process or it's just plain nasty. > Is that > in rc2 unstable? Not sure what Brett has working but I still see Jelly stuff in there. > Is it going to be rolled in anytime soon? Brett's going as fast as he can, and I will make one last attempt at fixing the problems in my branch as I can build anything with it there are just some glitches with in my handling of goals which is ultimately going to need to be fixed anyway I just want to do it after 1.0. > What is rc2 > unstable? Is it based on a completely new Maven core (plexus). No, neither branches are based on Plexus but I'm going to start bring pieces over bit by bit that are. > If so, > how could you merge it with head without breaking everything? With the branches I imagine we might have to do something empirical like setup DamageControl with a set of projects using the old and new code. Due to the nature of the code even with tests something will likely go awry. What I want to do is start using small components and Wagon will be the first one. Wagon is actually neither bound to Maven or Plexus but is a general artifact handling library which is small and currently stands at ~100% coverage. I also have small components for the core of Maven and they will also have close to ~100% coverage. It will be the only way to bring the code to some level of sanity. And these small bits of code can be introduced one at time as it is obvious that my massive rewrite of one blob with another blob didn't work. All artifact handling will be done with Wagon. So small artifact handlers can be written and tested so we can gradually removed the untested crap (which I take entire responsibility for in having not insisted on better testing practices) with components like Wagon which themselves are tested and are easier to test with. > Thanks > -Vincent (a bit confused :-)) > > > -----Original Message----- > > From: Jason van Zyl [mailto:[EMAIL PROTECTED] > > Sent: 17 November 2003 15:33 > > To: Maven Developers List > > Subject: RE: rc2 unstable > > > > On Sun, 2003-11-16 at 16:49, Brett Porter wrote: > > > Status update: I believe I now have all of the plugin context issues > > licked. > > > I haven't had a chance to fully test my last set of changes, but I > > > identified the problem. > > > > > > Jason, how did your work on the werkz code go on the weekend? > > > > I worked on Plexus and Wagon all weekend. I'm afraid to start on that > > Maven branch because it will inevitably mean I won't look at anything > > else for a span of days. > > > > But I think our work in the separate branches is primarily > complementary > > and originally comes from the branch I'm working on so whatever bits > and > > pieces go in don't really matter. I won't attempt to fix the werkz > > problem until I can replicate it the problem with a test. I am also > not > > averse to rolling in more changes post 1.0. If most of the memory leak > > is taken care I think everyone will be happy with that. > > > > > This leaves some very minor things left to fix and some more testing > > before > > > the RC2 branch is (hopefully!) fully functional again with all > existing > > > plugins/projects, unless they seriously abused the way the old > system > > > worked. > > > > > > >From there, I'd like to attempt to reimplement plugin caching (the > old > > > technique was incompatible) before RC2 is made available to reduce > the > > > startup time again. > > > > I did reimplement the searching and caching in a different plugin > > manager that had a number of tests and fixed the problem where the old > > plugin manager didn't find dependencies properly. It was a source file > > with a set of diagrams outlining the possible dependencies between > > goals, prereqs and other plugins. > > > > > Cheers, > > > Brett > > > > > > > -----Original Message----- > > > > From: Brett Porter > > > > Sent: Thursday, 13 November 2003 8:20 AM > > > > To: 'Maven Developers List' > > > > Subject: RE: rc2 unstable > > > > > > > > > > > > Done. I noticed you already did the plugins in CVS - I > > > > thought this was only on your copy. Cool. Much cleaner now. > > > > Will also make any merging later a helluva lot easier. > > > > > > > > Current status: the pluginVar tag is not working properly in > > > > a reactor, because project.context != context.project in some > > > > cases (it seems projects have multiple contexts). I need to > > > > draw this up and figure out where the double mapping crept in. > > > > > > > > This means touchstone runs regularly, but not via maven test > > > > in the base directory (eg bootstrap). > > > > > > > > Cheers, > > > > Brett > > > > > > -- > > jvz. > > > > Jason van Zyl > > [EMAIL PROTECTED] > > http://tambora.zenplex.org > > > > In short, man creates for himself a new religion of a rational > > and technical order to justify his work and to be justified in it. > > > > -- Jacques Ellul, The Technological Society > > > > > > --------------------------------------------------------------------- > > To unsubscribe, e-mail: [EMAIL PROTECTED] > > For additional commands, e-mail: [EMAIL PROTECTED] > > > > --------------------------------------------------------------------- > To unsubscribe, e-mail: [EMAIL PROTECTED] > For additional commands, e-mail: [EMAIL PROTECTED] -- jvz. Jason van Zyl [EMAIL PROTECTED] http://tambora.zenplex.org In short, man creates for himself a new religion of a rational and technical order to justify his work and to be justified in it. -- Jacques Ellul, The Technological Society --------------------------------------------------------------------- To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
