Moved: -> @shel@cybre.space is a user on icosahedron.website. You can follow them or interact with them if you have an account anywhere in the fediverse. If you don't, you can sign up here.
Moved: -> @shel@cybre.space @shel

just wanna re-iterate that if any people of color have requests for things the dev team can address to make Mastodon better for people of color then you always have an open eat with me and I will do my best to make it happen, so long as it is feasible within the OStatus protocol.

· Tusky · 21 · 20

@shel Yes, can we set up usernames using Unicode non-Roman characters?

That is: would صقرم@mastodon dot such and such be possible? How would that work with IDNs too.

Has anyone tested that out?

@mastodon @saqeram I'm bringing this to Eugen's attention. poor RTL support has been an issue in general.

@mastodon @saqeram okay so besides making a GitHub issue I've brought this up in discord, made a pitch, and linked to your post. implementing this is outside my technical expertise but I'm gonna push for it getting onto the task queue

@saqeram @mastodon Hi! So! Update. It looks like the OSTatus Spec only supports ASCII in usernames, so unfortunately unicode usernames doesn't seem feasible. That said we could definitely improve RTL support to help with unicode display names in languages like arabic and hebrew. So "Improve RTL support" is now on the queue. Sorry I couldn't get unicode usernames.

@shel @saqeram
Maybe using Punycode? صقرم would become "xn--wgbh3bn" when registering and it would be converted back when displayed. And if not using Punycode, there has to be a way, one was found for DNS names and emails so no reason not to find one for Mastodon :-)

@saqeram @ko hmmmm. I will look into it. In that sense the username is still ASCII but there's just something that looks at it and renders it on the screen as Unicode right? that's sort of what happens with the emoji short codes...

@shel @saqeram Yes, Punycode is what is used for (sub-)domain names with non-ASCII characters. For instance, the TLD .تونس
is converted to ".xn--pgbs0dh" when sending the request to a server (DNS, web, etc.)
My personal page is both available at kevin.beaubestre.fr and kévin.beaubestre.fr but in my server configuration, I had to write the later as "xn--kvin-bpa.beaubestre.fr", it's the web browser that handles the conversion.

@shel @ko Punycode is a workaround but it seems like a workaround for the ASCII legacy (1963), and it's 2017 so we should find a way that doesn't depend on ASCII workarounds. I may be misunderstanding the entire premise of Punycode but I remember being disappointed by all the xn-- and percent-encoding.

@saqeram @shel I agree that we should not have to work around ASCII but:
1/ we still want to follow OStatus specs so until we can have them updated, it's better than not having any non-ASCII char in username
2/ which encoding to use? A Unicode one, sure but there are UTF-8, UTF-16, UTF-32 and others...

@saqeram unfortunately we would break compatibility if we used Unicode usernames. It's gotta be ASCII but I'm seeing if I can get some of the new volunteers to work on punycode.