Adding a new tripple-click mode?

[madduck]

Hello,

I come from rxvt. One difference I discovered is that Kitty allows me to double-click on text, and depending on select_by_word_characters, a certain portion of the line gets selected.

If I triple-click instead, the whole line gets selected.

In rxvt, I could configure triple-click to actually extend the selection, and I had it ignore all control characters and extend to the next whitespace.

If I quadruple-clicked, then the whole line was selected.

Take for instance an ls output with a couple of funny filenames:

% touch {foo,bar,baz}·{blech,bla,bleep}
% ls 
bar·bla    bar·bleep  baz·blech  foo·bla    foo·bleep
bar·blech  baz·bla    baz·bleep  foo·blech

I cannot get the whole filename selected on any of these with double-click. After 20 years of rxvt, I naturally triple-click, but that selects the whole line.

How would you like it if a triple-click actually extended the selection to the next whitespace in both directions, and only a quadruple-click would select the whole line?

Obviously, this could be configurable…

[kovidgoyal]

Adding an extra click has an overhead, it means that after a triple click kitty has to wait before executing the action in case there is a fourth click. So, not a fan. That said actions taken on various mouse events are already customizable, see mouse_map in kitty.conf, but there is no quadruple press event. I will accept a PR to add a new select action to surrounding whitespace which you can then optionally map to triplepress that should ease your muscle memory issues and it's a fairly trivial PR as well. To get you started I have added scaffolding for it in my latest commit, you just need to implement the actual logic for it at line 788 in mouse.c

[madduck]

Thank you for your response.

There does not seem any delay when I do this with rxvt, as each additional click just extends the selection. So I think you would need to check if the third click is within time range of the second, and the fourth within range of the third, but none of that should introduce a lag.

I am hoping that the following screencast can illustrate the behaviour (sorry about the aspecyt:

Note how the selections extends to the whole line on third click if the double-click already selected to between whitespace characters.

I definitely use whole-line select, so I don't want to just replace triple-click. I am wondering if you are open to quadruple-click if it does not introduce a lag?

[kovidgoyal]

On Wed, Apr 23, 2025 at 01:14:03AM -0700, martin f. krafft wrote: Thank you for your response. There does not seem any delay when I do this with `rxvt`, as each additional click just extends the selection. So I think you would need to check if the third click is within time range of the second, and the fourth within range of the third, but none of that should introduce a lag.

Sure it will, when in range, which will be most of the time. If I want to do a triple click to select, I now have to wait half a second more before the action is triggered as kitty doesnt know if a fourth click will happen.

I am hoping that the following screencast can illustrate the behaviour (sorry about the aspecyt:  Note how the selections extends to the whole line on third click if the double-click already selected to between whitespace characters. I definitely use whole-line select, so I don't want to just replace triple-click. I am wondering if you are open to quadruple-click if it does not introduce a lag?

No, I'm afraid not, four clicks is a bridge too far.

[madduck]

Sure it will, when in range, which will be most of the time. If I want to do a triple click to select, I now have to wait half a second more before the action is triggered as kitty doesnt know if a fourth click will happen.

Maybe Rxvt just does it differently. I don't think it waits to see if there were 2, 3 or 4 clicks in total, and then does the selection; Instead — and I think this can be seen in the screencast above — it simply extends the selection on each successive click within time range. If I click twice and then paste within a millisecond, it pastes what has been selected. It does not need to wait to rule out a third click. If I click a third time, however, the selection is extended…

Maybe I am misunderstanding your concern. It seems straightforward to me, but you are the expert, for sure.

[kovidgoyal]

On Wed, Apr 23, 2025 at 06:28:28AM -0700, martin f. krafft wrote: > Sure it will, when in range, which will be most of the time. If I want to do a triple click to select, I now have to wait half a second more before the action is triggered as kitty doesnt know if a fourth click will happen. Maybe Rxvt just does it differently. I don't think it waits to see if there were 2, 3 or 4 clicks in total, and then does the selection; Instead — and I think this can be seen in the screencast above — it simply extends the selection on each successive click within time range. If I click twice and then paste within a millisecond, it pastes what has been selected. It does not need to wait to rule out a third click. If I click a third time, however, the selection is extended… Maybe I am misunderstanding your concern. It seems straightforward to me, but you are the expert, for sure.

In kitty mouse events can be arbitrary actions not just selections, therefore kitty has to wait to see which event occurred.