Is end_col exclusive? And if so it should be documented

[kmoschcau]

What the title says. I have run into this when writing a new builtin.

[jose-elias-alvarez]

MAIN.md specifies that null-ls diagnostic ranges are 1-indexed (and are later converted to the structure described in :help diagnostic-structure).

It's definitely not in the friendliest spot. I've wanted to add tutorials for writing sources but haven't had time to do so (#364 has been open for some time, but we haven't seen contributions).

[kmoschcau]

So the built-in help doesn't say either. From my testing it seems that it is exclusive. If I would set it like it was inclusive, the diagnostic highlight would stop one short most of the time.

[jose-elias-alvarez]

Exclusive or inclusive of what? The cursor? Thinking of this in terms of indexing makes more sense to me, and both our documentation and the Neovim documentation are clear on that subject (though I'd welcome PRs improving things on our end if you feel that they are confusing).

[kmoschcau]

No, what I mean is: Is the range represented by the start and end columns inclusive or exclusive of the end column?

This should be completely independent of whether columns are one or zero indexed.

For example, let's say we want to put a diagnostic on the example word "foo". The linter returns a start column of 4 and an end column of 6 for this:

123456789 // Column Numbers
fo foo ba // Example Words
   ^^^    // Expected Highlight
   ^^     // Actual Highlight

This is why I add 1 to the end column in the PMD builtin.