git-gutter:hide-gutter only respected if global-git-gutter-mode is active

[NateEag]

If you don't activate git-gutter by calling (global-git-gutter-mode), git-gutter:hide-gutter is not respected.

This is because of the implementation of git-gutter:reset-window-margin-p, which explicitly ignores the variable if global-git-gutter-mode is not active.

What is that logic intended to achieve? Is there a way of doing it that doesn't break git-gutter:hide-gutter?

[gonewest818]

At first glance, it seems the logic is "if the mode isn't active then why would you need to reset anything?" Can you explain a little further what it is you're trying to do?

[NateEag]

Sure.

I call git-gutter-mode in the hooks where I know I'll likely want it - prog-mode and text-mode.

That way I don't pay the cost of initializing it until I actually want to use it.

git-gutter:hide-gutter only working if the package is kicked off with global-git-gutter-mode and not when it's started by git-gutter-mode feels like a bug to me.

Why should one supported start mechanism work, but not the other?

[gonewest818]

Ok, now I see. Well, I went back through the logs and found using the global mode forces the gutters to be displayed always, and has been that way since at least 2013.

There used to be a setting called git-gutter:always-show-gutter that would also force the display when not using the global mode. This was replace by git-gutter:hide-gutter in this commit:
bffe95f#diff-0e0367812f9b0c9888d30842b9cbdee6L277-R280

And in that change it seems to me, the intentional choice was made to continue with the rule that the global mode forces the display of the gutter. So given that this package has gone unchanged for a while, I'm reluctant to break compatibility or change default behavior right out of the gate.

I agree it makes sense to optionally hide the gutter, so I'll look at ways to deprecate/transition to a new set of semantics without causing people too much disruption.

[NateEag]

Thanks!