Rendering of emoji presentation sequences

[fredizzimo]

I'm looking at this Neovide issue neovide/neovide#2990. I don't have any idea what the original issue is, but I discovered that we can't render simple emoji presentation sequences correctly https://www.unicode.org/emoji/charts/emoji-variants.html. And that may or may not be the issue.

For example from https://unicode.org/Public/emoji/latest/emoji-test.txt.

Both of these render the same for us

The clustering properly detects the emoji type https://docs.rs/swash/latest/swash/text/cluster/enum.Emoji.html, but CharCluster::map, returns the same char and glyph id. I'm not sure if they should be different or not, I don't know the specification well enough. And also, I don't see any special information that could be used in Shaper::shape_with either.

So, how are we supposed to deal emojis? Or is the swash support still incomplete?

For the reference, our current code is here https://github.com/neovide/neovide/blob/0f25975b8414e38c039b1be8173db59fb6cab7ae/src/renderer/fonts/caching_shaper.rs#L376

[xorgy]

Adding to this, there is also the issue of deciding override presentation and default presentation. For example, in a browser location bar typically the text presentation mode is forced even for characters marked with U+FE0F for emoji presentation.

The default presentation in many applications of U+2600 SUN is emoji presentation (for example, here on Github ☀), so 2600 FE0F and 2600 would both render like 2600 FE0F here, you would need 2600 FE0E to render the text presentation ☀︎.