Thanks Emmanuel for the thoughts. >> I tend to think that the balance between JIRA/MailingList is not easy to >> make. The problem is we seldom discuss specific questions deeply in JIRAs, thus we leave nothing context specifics for new contributors in future.
>>But I do think that ythe Mailing List get more attention than JIRAs. As said above, this is because we don’t favor JIRAs, we are used to the ML. I'm not sure it's all the good practice. Below just an example FYI. All the deep discussions can be tracked down here, from architecture, design to implementation. When some new guy wants to learn about the effort, we just point him/her there. https://issues.apache.org/jira/browse/HDFS-7285 Regards, Kai -----Original Message----- From: Emmanuel Lécharny [mailto:[email protected]] Sent: Friday, July 03, 2015 4:57 PM To: [email protected] Subject: Re: state of KDC Le 03/07/15 10:34, Kiran Ayyagari a écrit : > On Fri, Jul 3, 2015 at 4:24 PM, Zheng, Kai <[email protected]> wrote: > >> Thanks Jiajia for sorting this out! It's helpful to learn about the >> new project. Maybe we should have such status update or lighter ones >> regularly? >> >> > it is a good practice to let the team know before if you are planning > to add any new features or any _significant_ changes to the code base. I agree. Now, there are two aspects here : - about the reports : as a new project, and in order to keep the community informed about what is going on, yes, it's a good idea to post a quick report (once a month ?) - about the significant changes : I would strongly suggest they are discussed on the mailing list beforehand. JIRA is not the best place for that, because there are a lot fo them and it's hard to know which are related to minor issues and which are related to major refactoring or feature edition. I tend to think that the balance between JIRA/MailingList is not easy to make. But I do think that ythe Mailing List get more attention than JIRAs. One good exemple is the getIdentities() discussion, or the Cache discussion we have had lately. Those are the kind of threads that is very valuable for all of us. <about how The ASF works> More than that, the ASF is all about consensus building : that mean discussion, and if a consensus is made, then we go on. If no consensus appears, then we have two options (well, 3 but the third one is not something you want to experiment ;-) : o we vote, and depending on the result, we act. Kind of democratic, but it can lead to wrong decisions... Vote is more a way to assess a consensus. o we dismiss the proposal, and start thinking about alternatives. There is no one single way to do things. That are the two best ways to get things done withing hurting feelings. Let's speak a bit about the third possibility : o ego is hurt, people get angry, some call others names. At some point, the PMC has to take a position, and it may escalate to the Board, commit access is suspended or revoked, war ensue, atomic bombs are sent, the planet explode, the sun becomes a super nova, milky way collapse into a giant black hole, we vasnish into a singularity... It could even get worse ;-) More seriously, we are grown up adult (well, most of the time ;-), so we *dont't* want to switch in Solution 3 mode. That's the way we work ! </about how The ASF works> Ok, it was just a funny way to explain that discussin is always good !
