<C-A> to increase date will increase the year if cursor is not on the date

[kvalv]

Hey, this is a really nice project and I'm happy to see such active development on this. It already looks promising!

Describe the bug
Using <C-a> when the cursor is not on the date will increment the year, and thereafter it'll work properly.

To Reproduce
Steps to reproduce the behavior:

1. Write some text and then write :now: and exit insert mode.
2. Go to the start of line (e.g. by typing 0 (zero) )
3. Type <C-a>
4. The year has increased to 2022. Next time I type <C-a> the day increases, as expected.

Expected behavior
The date should increase by one day when I type <C-a>, even if the cursor is not on the date object.

Screenshots

simplescreenrecorder-.37.mp4

System information:

• OS: linux
• Neovim version/commit: Nvim: NVIM v0.5.0-dev+1449-gbdf3df402
• orgmode commit: 6b99d66

Additional context
<C-a> normally finds the closest number and increase it, so it makes sense that the first <C-a> increase the year (that's vim builtin behaviour). Subsequently the cursor is on the date object, and orgmode <C-a> functionality takes precedence and increase the day. However, I consider this a bug because the expected behaviour is that <C-a> should "do the right thing" and increase the date immediately.

I guess also <C-x> have the same issue.

[kvalv]

If you think it's worth fixing, I could actually try to fix it myself. I don't have much experience with lua, but this doesn't seem to be too hard to fix.

[kristijanhusak]

Idea is to have / increment/decrement date by one day when used within date context. Anything outside of it fallback to the default behavior.

As you said, Vim automatically finds first number and increases it. I don't think there's a way to prevent it from doing that.

Workaround is to bind date increment/decrement to another mapping, for example +/-:

require('orgmode').setup({
  mappings = {
    org = {
      org_increase_date = '+',
      org_decrease_date = '+'
    }
  }
})

That way / will work by default everywhere.

[kvalv]

Workaround is to bind date increment/decrement to another mapping, for example +/-:

I tried what you suggested, but it seemed to give the same behaviour as <C-a> / <C-x>, probably because the fallback is sent to nvim here: https://github.com/kristijanhusak/orgmode.nvim/blob/master/lua/orgmode/org/mappings.lua#L118

So I guess what I'm asking for is that <C-a> will first move to the date context (similar to how <C-a> is moving to the closest number) and then increase the date by one. Alternatively, it does not do anything if it's not on a date context. I think the current behaviour is not so intuitive, even though it makes sense by how vim's <C-a> and org-mode's <C-a> work

[kristijanhusak]

Oh, right, I need to fix that fallback to fallback to the defined mapping.

We could handle it by jumping to next number manually and then re-checking if it's date or not. Thanks for suggestion, I'll try to fix it.

[kristijanhusak]

@kvalv I pushed a fix for this.

1. Different mappings are now properly falling back to the ones defined
2. / Behaves as you suggested. It should properly jump to the date and increase it by a day.

Let me know if it works as expected.

[kvalv]

Hey, nice to hear you worked on this!

1. OK, so the way I understood it, if the cursor is on the beginning of the line and I hit <C-a> it should move to the next date object and increase it by 1 day. On the video, if I hit do that, nothing happens. (which is still much better than increasing the year)

2. Perhaps an unexpected side-effect of the change is that if I now type <C-a> without next dates, it moves to the nearest number (in this case 2), which I think also is not intended.

simplescreenrecorder-.39.mp4

[kristijanhusak]

Do you have default mappings set for this? It's not behaving same for you like it is for me. Here's gif how it works for me:

[kvalv]

Hey, yes you are right about that. I added this to the config:

 mappings = {
    org = {
      org_increase_date = '+',
      org_decrease_date = '+'
    }

and after removing that mapping it works as expected. Really nice, thank you very much! I guess this can be closed then.