Ooops. It was 190M when I was using 4 bytes per entry for lengths in

the dump file. The newer format is only 135M

p4raw-id: //depot/perl@23878

diff --git a/toke.c b/toke.c
index 369a2d7..6206c44 100644 (file)
--- a/toke.c
+++ b/toke.c
@@ -5585,7 +5585,7 @@ S_pending_ident(pTHX)
     return WORD;
 }
 
-/* Weights are the number of occurrences of that keyword in about 190M of
+/* Weights are the number of occurrences of that keyword in about 135M of
    input to Perl_keyword from a lot of real perl. This routine is about 20%
    faster than the routine it replaces.  */