On Thu, 2003-12-04 at 17:23, Brett Porter wrote: > We need to stop changing the roadmap and decide now. I thought we'd already > done that, but here goes again... > > The 3 options: > 1) release MAVEN 1.0 RC1 as Maven 1.0 final with any non-plugin-context > related bugs backported off the branch. Kill the branch and start 1.1 with > the new model. > 2) start integrating the new model into Maven 1.0 and kill the branch as > above. I don't think this is the best approach at this point. > 3) stick to the current plan of fixing the few remaining bugs in the Maven > 1.0 branch (about 6), and worry about everything else for 1.1 > > To elaborate on 3 - the reason it is floundering is because I am the sole > person working on it, and haven't had any time to do so recently as I was > moving house.
This is not directed at you exclusively. It is directed at my first attempt, Dion's subsequent attempt and your attempt. The problem with the attempts is that the core is so gory that collaboration is difficult at best. > I was knocking off about one bug a day until I got to the > current one which has had me a bit stumped. As of this morning I definitely > know what it is and just need to think how to fix it. Essentially it is that > plugin.context != context inside the plugin when running via the reactor, > the reason being that the goal tag has the context inside the plugin set the > first time and never changed. I need to switch this over at attainGoals(). > > Of the remaining 5 bugs I think 3 are already fixed (just haven't tested > them completely), and the other 2 are trivial (relating to differences in > output causing goal names and failure messages not being echoed at the right > time). > > After that, a lot of testing is needed with plugins, but that would be the > case with (2) above as well. > > Given that, I think (2) is out - its too risky in terms of breaking a bunch > of existing stuff. (100% test coverage means you can change it easily - but > doesn't say anything about how it corresponds to how things used to work > which is important for 1.0). > > This leaves us with a choice of releasing 1.0 now with the memory leak, or > later with the leak fixed, the code a bit more stable and reusable, and > project.properties inheritence working :) > > So lets vote and decide today because if we are not doing 3 I've wasted > enough time and won't spend a second longer on it, and will move on to > learning the 1.1 code. Welcome to my world. Dion tried to help but my code is too messy. I tried looking at your stuff and I find it equally hard to follow. It's the nature of the clusterfuck of the original code. > If it is (3), I'd appreciate someone else helping out, identifying the bugs > not already listed in maven-refactor-notes, and fixing things that don't > work as they used to. Until I sort out that current bug, reactor stuff is > probably best to wait. I honestly believe that two weeks from now I will be done the components. They are just so much simpler and we will more than likely get more help on those than the branches either of us have worked on. I approach I want to take is the one that is fastest in releasing 1.0 which I would say is releasing what we have and moving on to the new stuff. I am still going to push stuff from my branch into the release and I'm sure you will too but I don't think there is any reason in pursuing the branches anymore at this point. > 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]
