New magit-blame-quit micro-state stays on and breaks when you switch files.

[synic]

It doesn't turn off when you change buffers/files. After that, you cannot quit at all.

[syl20bnr]

This is a limitation of the micro-states, I don't see any satisfying solution to this problem. If we turn off the micro-state when switching the buffer then we have to quit magit-blame as well which is not very practical. We have the same issue with the time machine then.

Seems that a micro-state is not a good fit.

[synic]

Yeah, seems to be a problem with the scroll micro state too. SPC n . and then switching buffers screws it all up.

[StreakyCobra]

I don't know the underlying technology, so take my comment as an user point of view, and forgive my mistakes. As I just tested, it seems that a micro-state is specific to a buffer, and there can not be more than one micro-state active at the same time, even on different frames. So as soon as a buffer is changed, there is no reason to keep the micro-state open.

The question then become: Does it make sense to switch to another buffer when <micro-state-name> is active? If yes, it shouldn't be a micro-state. If no, it can be a micro-state.

For the scroll micro state, I don't really see reasons to switch to another buffer when scrolling, so it can be a micro-state. For the git blame and the time machine micro-states, users can want to switch to another buffer in some use-cases, but I think it is not the majority of cases.

So from my point of view, the better is to close a micro-state as soon as the buffer is changed, whether it is scroll, git blame or time machine. Also, git blame and time machine should be kept as micro-states because they are more useful in this form, but additionally other ways can be provided to use these functionalities in a way that allows buffers switching.

Here is my 2 cents.

[synic]

There's no use-case for switching buffers with the scroll micro state open, other than the rather large one of "didn't turn scroll micro-state off when I switched buffers", which is exactly what happened to me.

Once in this state you cannot turn it on for your new buffer. That, to me, is a broken feature.

[StreakyCobra]

So if micro-state gets always closed when switching buffers it should be fine, no?

[synic]

Yes, that would work.

[syl20bnr]

I pushed 7107ea5 which should fix this issue. The commit makes SPC g b idempotent so one can initiate it in a buffer, then switch to another buffer and discard the micro-state with q, when the user comes back to the magit-blame buffer he can press SPC g b again to reactivate the micro-state.

[synic]

Actually, now SPC g b does not enter a micro-state at all anymore, d'oh. Time machine still works, though.

[robbyoconnor]

Works for me

[synic]

Hrmm, seems to be working for me now too. I don't know what I did
differently.

On Mon, Oct 19, 2015 at 8:41 PM robbyoconnor notifications@github.com
wrote:

Works for me

—
Reply to this email directly or view it on GitHub
#3467 (comment)
.

[synic]

Should I close this?

[StreakyCobra]

When it's Fixed in develop, it will be closed at the next release. If you want to close it in the meantime it's up to you :-)