@CHAR and @UNICODE

[Charles Dye]

High time these two supported characters outside the BMP, i.e. 0x10000 <= n <= 0x10FFFF.

 

[rconn]

Since those characters won't fit in UTF-16, I'm not sure what you're suggesting. Did you want UTF-32 support or to do this with a 3 or 4 byte UTF8?

 

[Charles Dye]

Since those characters won't fit in UTF-16, I'm not sure what you're suggesting. Did you want UTF-32 support or to do this with a 3 or 4 byte UTF8?

They can be encoded in UTF-16. Characters above 0xFFFF are encoded as two wchar_ts, the first encoding the high ten bits of the character, and the second encoding the low one -- a "surrogate pair". See e.g. Wikipedia for the details.

So, when @CHAR finds a value above 0xFFFF, it should return the surrogate pair for the specified character. And conversely, when @UNICODE finds a surrogate pair in the input string, it should return a single value > 0xFFFF. (Values above 0x10FFFF are illegal, and should give an error message.)

 

[rconn]

Are you aware of any Windows Unicode font that supports code pairs?

 

[Charles Dye]

This is the current Take Command using Consolas:

But I don't know whether those characters are actually in the Consolas font, or whether Windows is just doing its font-substitution thing. I suspect the latter. Those glyphs look much the same in Lucida Console or Courier New.

 

[ClioCJS]

I would love to see this too.

 

[AnrDaemon]

Are you aware of any Windows Unicode font that supports code pairs?

Fonts support code points, code pairs is a way to represent a code point. And is not relevant to font per se.