my $utf8here, our $utf8here, and package variable $utf8here.

The actual minimal fix is in utf8.c and from NI-S,
the rest are the tests (in fresh_perl since I couldn't get
them easily to work elsewhere) and a slight behaviour change:
previously UTF-8 identifiers had to start with an alphabetic
character.  No more so, now they can start with an (Unicode)
ID_Continue character (which however is not a (Unicode) digit).
(Limiting the first character to ID_Start would be rather
restrictive, since ID_Start allows only alphabetic letters.)

TODO: use vars qw($utf8here).  This I don't find to be
a showstopper.

p4raw-id: //depot/perl@15943

1 files changed, 4 insertions, 0 deletions

diff --git a/perl.c b/perl.c
index b17448b279..30db9e32bd 100644
--- a/perl.c
+++ b/perl.c
@@ -686,6 +686,8 @@ perl_destruct(pTHXx)
     SvREFCNT_dec(PL_utf8_totitle);
     SvREFCNT_dec(PL_utf8_tolower);
     SvREFCNT_dec(PL_utf8_tofold);
+    SvREFCNT_dec(PL_utf8_idstart);
+    SvREFCNT_dec(PL_utf8_idcont);
     PL_utf8_alnum	= Nullsv;
     PL_utf8_alnumc	= Nullsv;
     PL_utf8_ascii	= Nullsv;
@@ -704,6 +706,8 @@ perl_destruct(pTHXx)
     PL_utf8_totitle	= Nullsv;
     PL_utf8_tolower	= Nullsv;
     PL_utf8_tofold	= Nullsv;
+    PL_utf8_idstart	= Nullsv;
+    PL_utf8_idcont	= Nullsv;
 
     if (!specialWARN(PL_compiling.cop_warnings))
 	SvREFCNT_dec(PL_compiling.cop_warnings);