Unicode limitation for usernames

[nukeador]

Hi there,

As part of the Community Discourse set up @Firefishy is leading, a question about unicode characters on usernames has been discussed and I was asked by @mmd-osm to bring this question here for guidance.

Discourse limits the characters on usernames to letters and numbers, but apparently we are getting usernames with unicode chars through the oauth2 connection.

We can enable unicode usernames over Discourse and we can also limit the allowed unicode characters, I suspect to avoid issues with invisible characters and confusing icons on the UI that might trick users:

What is the current unicode characters policy on OSM usernames? Is there an allowed list we can use too?

Thanks!

[tomhughes]

We allow almost anything - most of what we don't allow is actually ascii!

The canonical test user is @amandasaurus: https://www.openstreetmap.org/user/%E1%9A%9B%E1%9A%90%E1%9A%8B%E1%9A%90%E1%9A%85%E1%9A%87%E1%9A%90%E1%9A%9C%20%F0%9F%8F%B3%EF%B8%8F%E2%80%8D%F0%9F%8C%88

[tomhughes]

The main validation that is applied to usernames, other than length constraints and disallowing leading and trailing whitespace, is https://github.com/openstreetmap/openstreetmap-website/blob/master/app/validators/characters_validator.rb, which is applied with the url_safe option.

[nukeador]

Thanks @tomhughes

So I suspect we should be fine to accept any unicode, since OSM is already talking care of restricting, right?

[tomhughes]

I think so, yes.

Will discourse cope with a user changing their name? Is it using the numeric ID internally and will it update it's idea of name if the OSM site name changes?

[mmd-osm]

Yes, I think so. I've created user ᚐᚅᚇᚐ᚜ 🏳️‍🌈᚛ᚐᚋ -, logged on to Discourse, then changed the user name to ᚐᚅᚇᚐ᚜ 🏳️‍🌈᚛ᚐᚋ11 on osm.org, and logged on again using the new name on Discourse, and this all looked fine.

[amandasaurus]

It is fun to see that I'm the cause of this bug report (because the discourse software install had a problem when I tried to create an account), and used as an example of how OSM supports it all! 😂

Unicode, like human languages, is deep & complex. Every letter is “a unicode character”. If you plan to support it all, you may discover the edge cases of human writing 😈.

[mmd-osm]

The funny thing is, I couldn’t reproduce your Unicode issue and nothing had changed in the meantime 🧐

so right now it’s still not clear what your issue was…