Request to lemmy: can you please allow non-latin letters as well

submitted by Allah

I wish it was allowed to have persian letter usernames maybe even symbols as usernames it looks really cool and increases the username pool as well.

Log in to comment

26 Comments

Skull giver , edited

Deleted by author

SorteKanin , edited

ActivityPub users need to be identified by some identifier in the URL, and Lemmy chose the user name to be that identifier. As a result, non-Latin usernames becomeโ€ฆ complicated.

Sorry but this is just false. URIs can easily encode UTF-8 characters and it's perfectly standard to do so via percent-encoding. Example: https://en.wikipedia.org/wiki/๐Ÿ˜‚. Your browser will even automatically convert that ๐Ÿ˜‚ into the appropriate percent-encoding and will even display the emoji in the address bar, even if that is not the "true" URI.

This is, if you ask me, an unnecessary limitation in Lemmy.

Redjard

Link is detected without the emoji in my app. You might wanna hardcode the link as https://en.wikipedia.org/wiki/๐Ÿ˜‚
[https://en.wikipedia.org/wiki/๐Ÿ˜‚](https://en.wikipedia.org/wiki/๐Ÿ˜‚)

Asudox

Using ASCII in URLs is simple and is less error prone than "supporting" unicode via percent encoding. It is also just a convention to use ASCII for usernames in many platforms. ASCII is also supported out of the box in major OSes while some unicode characters might not. What about impersonation? And what about people trying to type in the username of someone that uses unicode? It is not logical to use unicode in this case.

SorteKanin

It is also just a convention to use ASCII for usernames in many platforms.

That's only true for platforms that only caters to the English speaking world. The fediverse should be and is much broader than that.

ASCII is also supported out of the box in major OSes while some unicode characters might not.

What? There is no major OS that does not support Unicode out of the box.

Percent encoding is perfectly fine and users won't even see it.

Also please stop down voting twice with your alt accounts, that's not cool.

sznowicki

Punycode would work here better I think as itโ€™s plain ASCI with no special characters except a dash if I recall correctly.

SorteKanin

Punycode is not solving the same problem. Punycode solves Unicode in *domain names*. Percent encoding is for Unicode in *URL paths*. Lemmy only needs to worry about the paths, Punycode should be "supported" out of the box without any special handling

Blaze

I see Arabic used from time to time

Allah [OP]

In usernames?

Blaze

Communities display names: https://lemmy.ca/c/maroc

I guess it could work for users display names too

AbouBenAdhem

This userโ€™s name is displayed in Arabic, although the characters in the URL are Latin.

RobotToaster

Looks like his *username* is in latin characters, but he has an arabic *display* name.

Allah [OP] , edited

Is it possible to make it in other than latin?

originalucifer , edited

have you ever seen a non-latin char url, ever?

the fedivere is incredibly url/dns dependent. labels/content can be any language (mbin uses weblate to allow for dozens off languages) but the underlying urls that control everything prolly require latin chars

Aatube , edited

https://ign.ไธญๅ›ฝ ? There's been a standard to encode it as xn-- for a while.

Allah [OP]

See this

https://ar.m.wikisource.org/wiki/ุชูุณูŠุฑ_ุงุจู†_ูƒุซูŠุฑ

intensely_human

Not from an ASCII

nutomic

I believe there is still an open issue on Github for this, but no one was interested to help implement and test it. So use the search function and contribute!

Asudox

You won't get non latin usernames anytime soon. But you can change the display name using non latin charactets

Bobby Turkalino

This thread is news to me. Unicode is Unicode, no? Why restrict to Latin letters?

Tanoh

There is also the risk of homograph attacks. The link below is for domain name encoding via IDN, but the same applies to usernames. You could easily impersonate another user by having chars that look similar.

https://en.wikipedia.org/wiki/IDN_homograph_attack

Asudox , edited

Because URLs are usually in ASCII. That was a standard. Check RFC 1738 and 3986. Now, you can use percent encoding, but why use that. It just complicates things.

SorteKanin , edited

There is a standard way to encode Unicode into URLs, it definitely doesn't have to be ascii. Percent encoding is used all over the place.

EDIT: I don't mind a down vote but double down voting me from your alt @Asudox@lemmy.world is not cool. That's sockpuppetry/vote manipulation.

Display Name field. You can use whatever you want. Even emojis. The feature is already in Lemmy; but not every instance has it available. Lemmy.World does use it, though.