terminfo.dev lists kitty as not supporting the kitty keyboard protocol, which is implausible

[egnor]

In https://terminfo.dev/kitty-extensions (and other places the datum is surfaced), the Kitty terminal (on mac) is listed as NOT supporting the kitty keyboard protocol. Kitty-on-Mac is the fourth column here:

Given that kitty originated this protocol it's somewhat hard to imagine it doesn't support it, though I wouldn't rule out bugs having crept in that might be at odds with your test suite. Even if there's some technical reason why this is correct, it would need to be annotated since it's so very surprising.

[kovidgoyal]

Eh the data on this site is embarrassingly wrong in so many ways. I just looked at the kitty page and the list of things it claims kitty does not support is utterly wrong. Some quickies XTVERSION XTGETTCAP are all supported. The keyboard protocol is supported, OSC 52 is supported OSC 5522 is supported (another protocol that kitty originated btw which this site fails to mention), color scheme reporting is supported etc, grapheme cluster cursor movement is not actually a thing in terminals etc.

[egnor]

yeah I was just wondering about that. from https://terminfo.dev/unicode/grapheme-cursor

Cursor movement (CUF/CUB) should respect grapheme cluster boundaries. Moving the cursor back over a ZWJ emoji sequence should skip the entire sequence, not individual codepoints. Many terminals still move by codepoint rather than grapheme cluster.

like you said my understanding is that no vtxxx-ish terminals move by either codepoint or by grapheme cluster, they move by character cell, and terminals stuff graphemes into one or two or more(?) cells ideally following your spec or something like it?

the test is presumably exercising some aspect of unicode processing that varies between implementations, but without digging deeper it's hard to say what would even be correct?

this site aspires to be a good resource, but suffers from too little human review methinks

[kovidgoyal]

Yes definitely, all VT cursor movement controls existed long before unicode and have absolutely no notion of grapheme clusters or moving by them. One should not retroactively change what they mean, they mean moving by cells only not by grapheme clusters or text content of those cells generally.