[seanmcdirmid]

Unicode sucks for this kind of thing, since it would go on disk. Even if they don't go on disk, you have to figure out how to break them up into multiple characters so they can be edited (so ⇒ must be two characters, which is messy). No, Unicode isn't the answer. It is the answer to deali with someone who isn't using a specific font, like readers on hackernews.

Ligatures are a cake and eat it to. To those who don't care for them, they simply don't see them. As they are under control of the font, editors don't need special support either.

[pdkl95]

> Unicode sucks for this kind of thing, since eh go on disk

It going to be hard to express the intent when programming if you never write that expression onto non-volatile storage.

> break them up into multiple characters so they can be edited

Several different methods exist for editing Unicode. (e.g. [1])

> As they are under control of the font, editors don't need special support either.

That's exactly the problem. Most editors used for programming already understand the full syntax for many programming languages, while font-ligatures only match short character sequences. As the article mention, this will incorrectly replace some things that happen to share the same sequence of characters. There are also problems[2] in editors with storage or drawing boundaries in the middle of a ligature. Mapping the correct characters to a replacement glyph is a lot easier when you understand the surrounding grammar.

[1] https://docs.perl6.org/language/unicode_entry

[2] The bug tracker for FiraCode has several reports of

[emn13]

The fact that they go on disk is a pro, not a con; and the fact that they're one character not multiple is also a pro and a not a con: what's the point of the individual characters? Those were merely introduces as a necessary workaround to cope with input limitations. Kind of like C trigraphs - and nobody uses those because they want to.

[MrQuincle]

Unicode is a disadvantage for someone who wants to contribute to your code. That person now has to figure out how to write down Unicode arrows, taking her/him out of her/his flow.

[kps]

Sure, and the Dane has to figure out how to write down ø, taking them out of their flow — until they learn to apply the correct tool for the job.

[MrQuincle]

I'm not sure if you're sarcastic or not. However, I'm Dutch, and we have similar tokens.

Nobody minds if you write Danish code, but if you like contributors, it is asking a lot of them to change their editor or tools to collaborate with you. I hope you didn't mean your collaborators have to adjust to you like that.

[kps]

Not at all; the point is that a Dane isn't actually taken out of their flow to type ø in Danish, because they're using a suitable keyboard layout. Likewise a programmer need not be taken out of their flow to type (a→t ≠ 0) or a ← b ∧ ¬c or (2=0+.=T∅.|T)/T←ιN.