On 10/4/18 9:23 AM, Paul.Koning@dell.com wrote:

I wonder if the Unicode standard says so, or if this is a Linux bug.

I looked into it, and it's arguably a bug in the glibc regular expression matcher: it does not recognize u̯ as alphabetic because it's composed of U+0075 U+032F and the latter is not alphabetic. Let's add a comment to that effect and remove ɪ from the macro since it works OK. Proposed further patch attached.