Re: [emacs-bidi] Re: Arabic support

Eli Zaretskii <[email protected]> writes:

> And another question: AFAIU, an LGSTRING specifies characters as
> Unicode codepoints, while the Windows Uniscribe APIs expect wchar_t
> wide characters, which on Windows means UTF-16.  This means we should
> encode the codepoints in LGSTRINGs to UTF-16 before passing them to
> Uniscribe, rather than passing them unaltered, right?  The current
> code will break for characters whose Unicode codepoints are beyond the
> BMP, right?

Theortically, yes. In practice I don't think any of the scripts that we
use the shaping engine for lie beyond the BMP.