Updates min/maxLenth tests to mention graphemes rather than unicode code points

[spacether]

Updates min/maxLenth tests to mention graphemes rather than unicode points Updates the language for the test that verifies that one grapheme is too short when checked against minLength of 2

[spacether]

This came up when working on my java implementation here: https://github.com/openapi-json-schema-tools/openapi-json-schema-generator and reading this article about grapheme length vs code points: http://illegalargumentexception.blogspot.com/2009/05/java-rough-guide-to-character-encoding.html#javaencoding_strnlen

[gregsdennis]

Also ref: https://json-schema.slack.com/archives/CT8QRGTK5/p1701324578397949

[Julian]

Looks good to me too, just change it in the other drafts too

[gregsdennis]

change it in the other drafts too

Oh! Please also check for maxLength tests as well.

[karenetheridge]

Maybe we should pick some characters that actually make sense -- as far as I can tell "\uD83D\uDCA9" isn't valid utf8.

I did some digging in the history and the test was added here: #52 - and indeed this is UTF-16, not UTF-8. we should fix that. The proper sequence for the poop emoji, that should appear in the file, is \u1F4A9 -- but I think we should pick something better -- a sequence of unicode codepoints that combine together into one grapheme.

[gregsdennis]

No, JSON requires that it be broken out into the surrogate pair.

To escape an extended character that is not in the Basic Multilingual Plane, the character is represented as a 12-character sequence, encoding the UTF-16 surrogate pair. So, for example, a string containing only the G clef character (U+1D11E) may be represented as "\uD834\uDD1E".

[karenetheridge]

aha, indeed - https://datatracker.ietf.org/doc/html/rfc4627#section-2.5

[jdesrosiers]

@spacether Just wanted to make sure you saw this request to add this change to other drafts as well.