It's a good initiative and sandbox is where experimental code should start.
Gentlemen, don't feel insulted or whatever, to discuss code, reviewers should see it and that's why sandbox. I'm very happy to see new codes in sandbox these days like OACC or Tomcat-Lite, good ideas should be shared for reviews and positive/negative feebacks and suggestions. Regards 2008/3/27, Rainer Jung <[EMAIL PROTECTED]>: > > Filip Hanik - Dev Lists wrote: > > jean-frederic clere wrote: > >> +++ CUT +++ > >> > >>>> All in all I feel accused by the above formulation. > >>> the answer to this all lies in the fact, that I didn't see an email > >>> on tomcat-dev about OACC creation, so if I missed it, my apologies, > >>> but I cant recall seeing one. > >> > >> That is in sandbox. sandbox is supposed to be a play-ground for every > >> committer, or has that changed? > > when one wants it to be an official release, then it should probably be > > discussed. > > > > Filip > > > I'm trying to reply to today's mails in one mail to keep it easier to > follow. > > Feeling insulted: Thanks for your answer. Fully accepted. I think it's > better to talk about emotions when they come up. > > In advance notice about OACC: there was no. It was long in my head, but > the work was actually done in one day. Was it a fault? Maybe. There are > the obvious Pros and Cons. I decided to show what I wanted to talk about > first, and then start a discussion, totally accepting, that the result > might not be the one I wish. > > Of course I knew, that there is a big potential for conflict and I > really was thinking about the subject for quite some time. At the end I > think there is no real reason for a conflict, because it is just one > more option. If the project decides not to offer this migration step, > because the risk of lowering adoption for the better newer cluster is to > high, I'm fine with that. > > Technical reasons: I really tried to carefully describe, that my > motivation does *not* come from any known technical problems with the > new cluster. I don't know of any, and the observations we discussed are > actually minor points (apart from JMX :)) ). I agree, that we should > simply work on those and they don't necessitate OACC. > > I'm really motivated by a very abstract argument: HA needs to work with > the lowest possible risk. New code always carries an increased risk. If > we have satisfied Tomcat cluster users which are not now starting a new > project but are having HA applications (yes I mean really mission > critical), at the moment they can only decide to be locked in with > Tomcat 5.5 or jump to the new implementation. If they want to move to TC > 6, that wish will mostly be driven by the web app developpers. The HA > stuff is mainly under project control of the operations people. Offering > a two step migration seemed to be a good thing to me. That's not talking > about an existing customer, I get these ideas from my consulting > experience. > > If OACC is to much work: then no. > If we think that we can't get the story clear for our users: then no. > > But I think it is very little work, and I will accept any formulation > about our recommendations. > > Formally for me it is only a sandbox right now. I added the original > docs with some additions about the OACC situation and I need to add the > tar and zip targets. Finally I have one patch for DeltaManager, which is > also missing in TC 5.5 (and maybe HA/Tribes, I need to check). That's it. > > Then I would hope, that the project has a look at the stuff, and at the > final wording in the docs (I might formulate even stronger than I did > when I wrote the files yesterday). > > After that I would ask about existing formal limitations for placing > downloads on t.a.o/dev/dist, or p.a.o. > > Finally I would like to start a discussion, if we think we should have > it as an extras download, or not. > > Depending on the review, the discussion and the vote it will either get > released, or simply stay in the sandbox. I'm totally aware of the fact, > that this might happen. > > If you agree, we can postpone the OACC discussion for now. I don't plan > to put more efforts into it than maybe another hour. And we'll resume > the discussion in a few weeks, before we need to decide about releasing > it or not. > > But if you like, we can start another discussion about the remaining > minor HA/Tribes points. > > Regards, > > > Rainer > > > > > --------------------------------------------------------------------- > To unsubscribe, e-mail: [EMAIL PROTECTED] > For additional commands, e-mail: [EMAIL PROTECTED] > >
