[AD] unicode.c

[neil@xxxxxxxxxx (Neil Townsend)]

I am plaing around trying to get 3.9.21 wokrin on a linux/X box - not sure
if it should or not, the docs are not clear and I have to confess to not
following that side of the discussion. However, I have noted that line 312
of unicode.c is as follows:

   int c = *((unsigned char *)((*s)++));

this appears to cause a seg fault a bit later in the code that calls
utf8_getx. My observation is that changing the line to:

   int c = (int)(*((unsigned char *)((*s)++)));

solves the problem, although I am not convinced that the compiler shouldn't
be doing that automatically.

Neil