No attitude was intended. I was typing on my phone in the doctor's waiting room so I tried to be concise. Apologies if it came across badly.
I stand by what I said, though. I don't think the problem is Gerrit (not that Gerrit doesn't have its fair share of problems). The problem is having unrestricted write access to the canonical repository, compounded by rules that prevent fixing mistakes when they are inevitably made. And my best suggestion for complying with these ASF infra rules while maintaining a safe and sane working method is to have devs, as a day-to-day standard process, set their upstream "origin" remote be set to the read-only Github mirror. The discussion of whether Gerrit can fit into the ASF rules at all is, I think, a completely orthogonal discussion, although both discussions need to reach a conclusion before we can all be comfortably working again. There is a separate thread regarding Gerrit already, so I'd suggest continuing that discussion there. Ceej aka Chris Hillery On Tue, Jul 14, 2015 at 8:30 PM, Henry Saputra <[email protected]> wrote: > Chris, > > Let's figure out the best way to do it while adhering to ASF infra rules. > > I understand things are different in ASF that from outside but we need > to figure out the right rather than either my way or none attitude. > > - Henry > > On Mon, Jul 13, 2015 at 3:48 PM, Chris Hillery <[email protected]> wrote: > > That is correct, and that doesn't need fixing - it is the correct layout > > for us. > > > > What needs fixing is finding a way to prevent accidentally pushing to the > > wrong git remote. Unfortunately the ASF repositories don't offer any > > permissions or other mechanisms to help; in fact they require us to set > up > > the "wrong" way. > > > > My recommendation would be for all devs to pull from the github ASF > mirror > > and push to Gerrit, and not have a remote for ASF at all. > > > > Ceej > > aka Chris Hillery > >
