Comment by geocar | Hacker News Reader

Comment by geocar | Hacker News Reader

Parent Full thread

> My heuristic is something I'll call SMS sniff test: would it make sense to type that symbol into a text message? If it wouldn't, then it doesn't belong in Unicode.

I accept that heuristic, and find myself agreeing with it for the purposes of the external link character. Let me try to convince you:

This is a real conversation:

A> Are you familiar with Homebrew? You can get it from brew.sh.

B> Ok. Where is brew.sh?

If we were not limited by plain text, we could use colour and underlining to distinguish that as a domain name. A lot of websites find this poor for accessibility reasons so they use the link character.

Another way to think about this is to imagine the link-character is pronounced "https://" as in:

* Are you familiar with Homebrew? You can get it from https://brew.sh.

* Are you familiar with Homebrew? You can get it from {link character} brew.sh.

Isn't that plain text? I'm not sure how to pronounce it, but there's a lot of emoji I don't know how to pronounce. Do you think this is important?

So if you get that far, how much further is this really?

* Are you familiar with Homebrew? You can get it from brew.sh {link character}.

Historically <> were used; what happened to that convention? See <http://example.com/> or email me at <root@localhost> for details.

I think it's been conflated in online messaging in the last decade or so as people were imitating markup tags to emote. Eg:

<sarcasm>Oh yeah, that's right</sarcasm>

There is no squiggle that I can not hypothesize some way of using in a text message. The question is more, is there already some existing demand for it, before you asked the question? To which the answer would seem to be "no".

That's a contingent answer. If you, say, set out to try to convince the world to use it that way, and created that demand, then by golly, that demand would exist at that point and the answer could change. But hypothesizing some possible text message that could use it isn't strong enough to argue for inclusion in the standard because every possible proposal passes that test.

That sounds like an issue that could be solved by updating the renderer's gtld list with the various new ones that have been coming out instead of adding invisible meta characters.

Phones already do it for .com, .gov, etc, but our cutesy .sh, .dev, .rocks, .xyz, etc will take a bit to catch up.

> That sounds like an issue that could be solved by updating the renderer's gtld list with the various new ones that have been coming out instead of adding invisible meta characters.

Negative. My renderer is a piece of paper.

... Why would an invisible meta-character for links be necessary if the output is paper? You can already specify a link by prepending 'http://', etc.

Sorry. I thought that would make it clear.

The request is for a code point to represent the visible external link character, not for an invisible control-code to decorate some structured data (which cannot "appear" on paper).

If your renderer is a piece of paper, how is Unicode relevant at all? You’re free to freehand anything you’d like.

Writing out the "https://" still seems preferable, as long as http continues to exist and bare URLs try it by default.

(The hsts preload list helps a bit with this, but only for domains registered with it.)

website, or site, are the words you are looking for.

Which also lack a decent Unicode symbol despite being a word used in plain text. ️

Reply on news.ycombinator.com