Thread overview
Please don't force-push to shared branches
Dec 10, 2016
David Nadlinger
Dec 11, 2016
Seb
Dec 11, 2016
kinke
December 10, 2016
Dear all,

The release-1.1.0 branch was force-pushed to recently. As a hard and fast rule, don't do this on any branch you can reasonably expect other people have checked out as well. There are obviously exceptions (emergency fixups after committing big files, breaking the history, etc.), but in such cases it would still be considerate to then notify other people of what is going on. This is common open-source folklore for a reason.

Thanks.

 — David
December 11, 2016
On Saturday, 10 December 2016 at 17:03:51 UTC, David Nadlinger wrote:
> Dear all,
>
> The release-1.1.0 branch was force-pushed to recently. As a hard and fast rule, don't do this on any branch you can reasonably expect other people have checked out as well. There are obviously exceptions (emergency fixups after committing big files, breaking the history, etc.), but in such cases it would still be considerate to then notify other people of what is going on. This is common open-source folklore for a reason.
>
> Thanks.
>
>  — David

How about enabling force-push protection on GitHub?

https://github.com/blog/2051-protected-branches-and-required-status-checks
December 11, 2016
On Saturday, 10 December 2016 at 17:03:51 UTC, David Nadlinger wrote:
> Dear all,
>
> The release-1.1.0 branch was force-pushed to recently. As a hard and fast rule, don't do this on any branch you can reasonably expect other people have checked out as well. There are obviously exceptions (emergency fixups after committing big files, breaking the history, etc.), but in such cases it would still be considerate to then notify other people of what is going on. This is common open-source folklore for a reason.
>
> Thanks.
>
>  — David

That was me obviously, and the reason is the same as for beta4 - keeping the history (and branch comparisons) clean by rebasing the 3 1.1-specific commits (reverts+backport) onto master (and getting rid of your 2 then redundant cherry-picks for beta5); there were are about 30 commits since beta5, and neither cherry-picking nor merging master made sense to me. I've had a discussion with Johan starting at https://github.com/ldc-developers/ldc/issues/1862#issuecomment-263003598, your input there would have been more helpful than this forum post.