[Development] [FYI] new git-gpush features, a.k.a. the smart way of pushing to gerrit

Ulf Hermann ulf.hermann at digia.com
Fri Oct 24 14:34:48 CEST 2014

Another idea I came up with last time I accidentally rebased something:

Sometimes you will make a mistake and overwrite a change you didn't want to overwrite. No script in the world can save you from that. After that has happened we can basically say one of two things:

a, You should be punished. Go push another change and be very nice to the reviewers (and possibly, the original author). Then, maybe, after their annoyance has worn off, it will be approved.

b, You should be able to undo your push. Technically it shouldn't be that hard to roll back the change to the previous version, restoring the previous reviews, dependencies, and author. That will create some additional noise but it won't require any action from anyone but the person who made the mistake.

Of course, if b is implemented, that might increase the chances of that happening in the first place as the author isn't punished so badly anymore for making that mistake. Thus, people might disagree. I'd say it still makes sense because it would probably result in more changes getting in with less work from everybody.


More information about the Development mailing list