Unintentional double commits

[Ryan Schmidt]

On Dec 21, 2016, at 04:41, Andrea D'Amore wrote:

> While trying to push a small change (shell/xonsh) I managed to rebase
> AND merge the about 300 commits since my previous update,
> as a results those 300 commits in master now are "duplicated".
> 
> The files in master are unaltered, except for the actual portfile I
> was pushing, but the history is now messier.

As you know, I don't understand git, so I don't understand what happened or how it happened or what the implications are.

You said master is unaltered, but the summary message says the changes were pushed to master:

> Andrea D'Amore (anddam) pushed a change to branch master
> in repository macports-ports.
> 
>    from 1d95493  llvm*: Remove $Id$
>     new 7b98c6f  port py-virtualenv: add note about default selection
>     new 87373c8  crossbinutils-1.0: move libiberty header
>     new 82697c5  crossbinutils-1.0.tcl: trivial doc changes

I don't understand how the changes could be pushed to master when master already contained those changes. And if they didn't go to master, where did they go?


> I've been checking on IRC how to address this for last two hours, I
> had the dilemma that force pushing to the commit before mine would
> clean the history but at the same time break workflow for users that
> may already have fetched the changes.

We prevent force-pushes to master and release branches. I have no idea how many things would break if we allowed force pushing, and I'm not eager to find out.