14

I was intending to test out how to ping someone with a non-ASCII username - as part of the response to this post.

Since it specifically spoke about non-English languages, and frankly, it amused me, I tried to change the username of my alternate account from aibobot to ஐபோபோட். It's the same word transliterated to the Tamil script. We're the only language that uses it. Now I'm a little distracted and slightly confused as to why these are not accepted as letters.

Now, I can confirm that the letter ஐ is an actual letter. I wonder if the போ's are tripping it up (they are compound letters and have ligatures and other fancy typography), or maybe even the ட் is the problem. It tells me however that:

Display Name can only contain letters, digits, spaces, apostrophes or hyphens and must start with a letter or digit

Am I tripping up some less than obvious rule, or something put up due to abuse or spamming?

Full error message and example screenshot below.

Screenshot showing the error message of 'Display Name can only contain letters, digits, spaces, apostrophes or hyphens and must start with a letter or digit' in red text which is given when attempting to change your Display Name to one the system won't accept

Improve this question
6
  • 7
    Oh hey, you've nerd sniped me and I'm answering this outside of my normal hours. I did recently write some validation to check for invisible characters, but I don't think this is what you're hitting. I remember seeing some other validation that blocked certain character sets from previous malicious display names, and the Tamil Unicode character set sounds familiar, so maybe some of these are blocked as part of some older spam prevention?
    – Kyle Pollard StaffMod
    Jan 14 at 7:50
  • That crossed my mind - though I figured most of it was CJK. And its a fairly uncommon character set - single language (though I didn't try the other indian languages - this is the one I read). Also I literally asked cause I was nerd sniped too
    – Journeyman Geek
    Jan 14 at 8:01
  • 6
    Ah, yeah someone probably used an invisible character from it to hide their username a long time ago in a Stack Exchange far far away, and the rule hung around ever since. I'll mark this as [status-review], and one day I'll try to put your specific case in my gigantic display name unit test to see if it's fixable. No promises on any timeline :)
    – Kyle Pollard StaffMod
    Jan 14 at 8:12
  • 1
    Hypothesis: Only \w (word) characters are accepted, in addition to some punctuation found in western names (hyphen, apostrophe, space, underscore). Also, see this question about Devanagari characters
    – Laurel
    Jan 14 at 14:15
  • hm - I guess the (probably wrong term) composite letters like the போ and the consonants like ட் use diacritics for the modifiers - so that might be it
    – Journeyman Geek
    Jan 14 at 14:20
  • 3
    @Laurel I don't believe we have a \w regex validating things here, I believe there were just certain ranges of characters we didn't allow (again, for some older spam prevention)
    – Kyle Pollard StaffMod
    Jan 14 at 19:58

0

Reset to default

You must log in to answer this question.

Browse other questions tagged .