On 23.01.2021 21:02, Lars Ingebrigtsen wrote: > Dmitry Gutov writes: > >> We're talking about slightly different things. >> >> When I hear "mouse-2 binding", I think the global bidning in >> vc-dir-mode-map, which affects how mouse-2 works also when clicking >> outside of any buttons (on whitespace beside the buttons). And I think >> that behavior is relatively weird and could be removed. > > I didn't even notice that `mouse-2' did stuff when outside the > "buttons". And what it does is quite odd... hm, no, if you click after > the final line, it'll mark the last line, and before the first, it'll > mark the first, so perhaps that's logical? I meant more like between the buttons, but on the same line as them. The "before" and "after" behaviors are logical, but the whole binding is very non-discoverable. Nor does it work on file names. >> But we also have the behavior of mouse-1 and mouse-2 clicks on the >> "status" button. Which seem fairly sensible to me, but might be the >> cause of Glenn's annoyance in this report because it's relatively easy >> to click mouse-1 anywhere in the buffer by mistake. > > Yes, I think he was complaining about the "status" button. > >> Right. But if that is what is at issue here, I think the question >> becomes whether to remove the "buttons" from that column (and keep the >> mouse-2 binding globally) or whether to remove the mouse-2 binding, >> but set up the buttons to continue working. Or make mouse-1 stop >> working on the buttons but keep mouse-2 working everywhere (except the >> file names, I guess?). > > I think removing the global mouse bindings and only having them on the > "buttons" would be least surprising -- that's how most modes work. The attached patch seems to do that, except for the "status" buttons beside directory names, which seem to be printed by ewoc in some other way, and thus don't go through vc-dir-printer. Which is a new different kind of awkward. We're also solving a different problem than the one which was reported. So I'm not sure about the change, TBH.