Michael Seibt

> It should close if: > cursor changes position by arrows, pgdn/pgup The up/down arrows (and pgdn/pgup) are needed to select the desired auto-completion entry.

> Reverting [f061ce5](https://github.com/gitextensions/gitextensions/commit/f061ce5443eb5f784f40ed9f51cf1a68f7a4f280) According to [#8851 (comment)](https://github.com/gitextensions/gitextensions/pull/8851#discussion_r575733576) that should not have an effect. Furthermore, `ThreadHelper.JoinableTaskFactory.Run` would still have blocked the UI thread. The entire `DisplayAheadBehindInformation` could be made `async`. ---...

> This was changed partly by your comments as the old value was misleading. The commit button is grey in this situation. This is true only if GE is minimized...

I have found how to reproduce this or a related issue. If git processes are terminated on cancellation request, the status of the artificial commits is not updated. The issue...

> If a remote has too many branches (especially the trunk) then it is more a business/team discipline issue, and perhaps regular pruning and forks should be considered. ... As...

Updating Git was blocked either - until I closed VS Code. (I had closed all GE instances before). Since the Git update, the de/activation of remotes works again. Though this...

My 2ct: We are not forced to follow the recommendation of double gutters for windows larger than 640 pixels, which would mean nearly all.

`Consider the possibility "FormBorderStyle = None".` Only if dialogs stay resizeable!!

> @mstv any further concerns? Yes, as replied to a code comment above: With the current approach, the user must reopen the settings anyway because - due to the swallowed...

In addition to the flaky tests on AppVeyor, the different integration test `RunCommandBasedOnArgument_history("blamehistory",True,True)` failed locally once - with a suspicious error. This means the cancellation should be awaited somehow. Or...