On 16/12/2011 20:21, Antonio Petrelli wrote: > 2011/12/16 Mark Thomas <[email protected]> > >> On 16/12/2011 16:06, Antonio Petrelli wrote: >>> Please Mark calm down. >> >> No, I will not calm down. The release process has been changed without >> prior discussion and on top of that it is now broken for Maven >> artefacts. That is not acceptable. >> > > > Mark, everything's right,
No it isn't. If everything was alright, there would have been no need to start this thread. > good boy :-D I'm going to ignore that patronizing remark. > First of all, did you try to release something and it went wrong? No I didn't. However, as a member of the ASF infrastructure team I know enough about how Nexus is configured to know that due to [1] releasing Maven artefacts via scp+rsync using people.a.o is now blocked. All of the Tomcat build scripts are written to release JARs files to Maven Central via scp+rsync via people.a.o. Because of the switch to Nexus, the publishing to Maven Central part of the build process is currently broken. > If not, remember that changes might be rolled back at any time (though, > sincerely, I don't see why, especially if you did not see a problem). This is not a simple change to a build script in svn. Ultimately anything can be reverted but reverting the switch to Nexus looks to be non-trivial. As an example, the entire tomcat section of the repo on people.a.o has been removed and would have to be reconstructed. That is far more work than a simple svn revert. I am sure that other changes have also been made. I do not know how much reverting all of them may entail. To re-iterate I do see a problem. The build scripts are all written to use scp+rsync via people.a.o but releases via that route is now blocked. The good news is that Jean-Frederic has indicated via private e-mail that he will be fixing the build scripts to use Nexus tomorrow. I would like to see him confirm that on this list but I have no reason to doubt his word. A quick read of [2] suggests that it should be do-able in that timeframe. > And anyway, your attitude against Jean Frederic is unacceptable, in this > cases you should remind that: There we disagree. I think that Jean-Frederic's decision to go ahead with [1] without community discussion was the wrong decision. That the release of Tomcat artefacts to Maven Central is now broken makes things worse. > * Apache is a community driven foundation, Exactly. Which is why one individual making technical decisions that affect a project's release process without prior community discussion is wrong. To repeat what has been said many times before at the ASF "If it didn't happen on the mailing list then it didn't happen" and there was no discussion of this change on the mailing list. If the discussion had taken place on the dev list, the issues around the current build scripts would have been identified and addressed and, for example, we could have co-ordinated the switch to Nexus with the necessary changes to the build scripts. And if we hadn't identified the issues then this would be a community screw-up that we could all share responsibility for. You'll note that I have started a separate thread to discuss the relative merits of Nexus vs scp+rsync. If you are aware of anything that would indicate one is a better choice than the other for Tomcat then please do contribute to that thread. > not a private firm that will lose money if the build does not work. I agree this has nothing to do with money. It has everything to do with community. Part of the Tomcat community relies on obtaining the Tomcat JARs from Maven Central. Based on how quickly questions get asked if I forget to run the Maven artefacts part of the release process once a Tomcat 7 release vote has passed, I would guess that a sizable part of our community depends on those JARs. A direct consequence of this change is that we are currently unable to make releases to that part of our community. Currently we have no plans for a release but it only takes one critical security vulnerability report to change that. > * I think that you should recall the basic rules of personal relationship. > Jean Frederic is a person, attacking him this way is *bad*. If someone screws up, I am going to call them on it. If I screw-up (and I do regularly - check the archives for the long list of stuff I have managed to break over the years) I expect to get called on it. Doing things without any on-list discussion is a huge no-no at the ASF. I would agree my response was a strong one but I wouldn't characterise it as a personal attack (Jean-Frederic if you read any of my comments were read that way I apologise - that wasn't my intention). > He probably made a mistake, however there are many ways to tell things, most > of them > are kind. Everything is relative. I would say this thread was pretty mild compared to others I have seen both on this list an elsewhere at the ASF. Could it have been milder? Sure. But when I see such a fundamental part of "The Apache Way" - all technical decisions are made on list - being forgotten it is very difficult not to react strongly. Believe it or not, the first draft of my e-mail that started this thread was a lot stronger and threw around phrases like "-1 veto" that on reflection I decided to tone down. Mark > > Best regards > Antonio > [1] https://issues.apache.org/jira/browse/INFRA-4162 [2] http://www.apache.org/dev/publishing-maven-artifacts.html --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
