[Flex-devel] unicode unsigned short

[Paul <pa...@pr...>]

The unicode16 version of flex uses unsigned short.
Without opening a large can of worms re types, to pass the tests utest-posix
and utest-posixly-correct (nothing to do with posix compatibility) with
the -U flag on, I have altered the line:

char * tests[NUM_TESTS] =  { "ababab"}; /* non unicode */
to (simplified):
unsigned short testu[7]={'a','b','a','b','a','b',0}; /* altered for 
unicode pn */

The test then passes with testu OK.

This makes initializing a pain since:
a.    wchar_t is 4 bytes wide in gcc
b.    L"ababab" is also 4 bytes wide.
whereas unsigned short & unicode 16 is 2 bytes wide.
I will note that char16_t is not defined in C, only C++.

Thoughts please.

Paul