https://bugs.exim.org/show_bug.cgi?id=1719
Philip Hazel <ph10@???> changed:
What |Removed |Added
----------------------------------------------------------------------------
Resolution|--- |FIXED
Status|NEW |RESOLVED
--- Comment #2 from Philip Hazel <ph10@???> ---
I did find a short patch that fixes this, so it will be in 8.38. The bug was
that when [:^ascii:] or [:^xdigit:] or (in the case of PCRE1) [:^cntrl:] were
in a class with a Unicode property (and PCRE_UCP with \w turns into a Unicode
property), the fact that all wide characters were in the class was not being
recognized.
While checking this out I noticed that PCRE and Perl handle \w differently in
Unicode modes. PCRE documents that it converts to "any character that matches
\p{L} or \p{N}, plus underscore" but Perl also includes "combining characters".
PCRE has been this way since PCRE_UCP was introduced in 2010.
--
You are receiving this mail because:
You are on the CC list for the bug.
