Yep, you are right, it is actually "c:\wïñëtèst". I will fix it in the next try.
That won't work. The actual unicode characters have different codes. Personally I don't see why you need non-acsii characters to test unicode APIs.
In this case, the non-ascii characters are necesary to expose the bug since the bug is enclosed in -- snip -- if (*c<=0x20 || *c>=0x7f || strchr(unsafechars, *c)) ... -- snip -- Regular ascii chars will not enter the buggy block.
to the top of you're test will do the trick in this particular case. Btw. use skip instead of trace for the above.
Cheers,
Paul.
Just saw Dmitry's comment. Another thing you can check is the lasterror (GetLastError()) as some of the W-calls are present but will return ERROR_CALL_NOT_IMPLEMENTED when asked for last error.
Thanks, I see how it is being done in user32 tests. I'll change the code to check for ERROR_CALL_NOT_IMPLEMENTED after making a W-call.