Hi
The encoding of six bytes long utf-8 sequences is buggy. Today unicode
requires at most 4 bytes long utf-8 sequences but if we handle 5 and 6 too
then let's do it the right way.
The attached patch was created using a fresh master clone
(d894cfd104086ddf68c286e67a5fb2e02eb43b7b).
I'm writing a tool (pxargs, an xargs variant) that can accept input strings
in shell-quoted format and used the bash manual and source code as
references. I haven't compiled the latest bash sources to check the bug but
it is likely to affect ANSI-C quoted strings like $'\U7fffffff'.
Best Regards,
Istvan Pasztor