Dot-repeat in visual mode

[tvsfx]

I usually use nvim-surround in visual mode, but noticed that dot-repeat does not work there. This is confirmed in the Dev Wiki. Is there any particular reason why this might be the case? Is it because the motions used in visual mode are more difficult to retrieve somehow, which prevents your trick (https://gist.github.com/kylechui/a5c1258cd2d86755f97b10fc921315c3) from working? Is there some workaround to make it work? Thanks in advance!

[kylechui]

It's because once you select something in visual mode, there is no concept of the motion used to create the selection (since you might have used many motions many times, e.g. iwwll). Thus we only really have start/end coordinates available, which isn't particularly helpful for dot-repeating.

[tvsfx]

Yes, that's more or less what I expected. However, dot-repeat for e.g. change or delete in visual mode does work, so it looks like nvim manages to track and replay all of the motions that the user performed before a change/delete. Is this a limitation of the way you implement dot-repeat?

[kylechui]

Based on testing on my own machine, it doesn't work. It seems to change/delete the same number of characters, but does not respect the motion that generated that selection.

[tvsfx]

You're right, I thought I tested for that, but I must've coincidentally picked words of similar length. My bad.

So does that mean you could theoretically still dot-surround words of the same size? That could be useful to surround the same selection multiple times, outside-in.
In any case, I guess I'll switch to surrounding without visual mode then, that seems more convenient. Thanks!

[kylechui]

Current implementation doesn't support dot-repeating in visual mode at all; I imagine 99.9% of use cases revolve around surround semantic text objects, rather than exactly k characters or something to that effect.