> On 8 May 2019, at 10:24, Liang Qi <[email protected]> wrote: > > > >> On 7 May 2019, at 22:13, Oswald Buddenhagen <[email protected]> >> wrote: >> >> On Tue, May 07, 2019 at 01:52:48PM +0000, Liang Qi wrote: >>> This is a very common issue when I do the integraton work in qt5. >>> >>> For example, currently we have 5.12->5.13->dev merge order, and the last >>> round 5.13->5.13.0 today. One change(changeA) had been integrated in 5.13, >>> and another change(changeB) with duplicate Change-Id was pushed to 5.13.0 >>> branch with status:open. When I try to merge 5.13 to 5.13.0, the changeB >>> will be picked up by gerrit and be part of the integration, which I found >>> in COIN. Normally that integation will fail, but I am not 100% sure. >>> >>> I only know the phenomenon, but not the root cause. >>> >>> If you have a change was merged in lower branch, please try to avoid use >>> duplicate Change-Id in upper branch, at least in the current merge order. >>> >> please don't give such bad advice. > > It’s not advice, more strict than that. This situation had wasted quite some > time and energy from me and CI/COIN. > > Here is just the latest case which happened yesterday. > > When I tried to do the final down merge qt5 5.13->5.13.0 plus the submodule > update in 5.13.0, after staged, I got > https://codereview.qt-project.org/#/c/260927/ included in the integration in > COIN. I need to cancel it and ask the owner or gerrit admin to abandon it. > That takes too much time, especially if I tried it during midnight…
I don’t understand why a change that is just sitting there on Gerrit can affect the merge. Nobody ever staged that one onto the “wrong” branch, right? And then the bot detected the duplication and auto-abandoned it. _______________________________________________ Development mailing list [email protected] https://lists.qt-project.org/listinfo/development
