[haru-s]

Hi!

The LPSTR is defined by "alias char* LPSTR".
But D's char is a utf-8 string, so LPSTR must be defined by "alias ubyte* LPSTR".
This problem is broadly applicable to other points of the Bindings.

[doob]

That would cause a lot of casting between char* and ubyte* which would be very annoying and since all valid ASCII is valid UTF-8 I don't think it's any problem. "The first 128 characters of the Unicode character set (which correspond directly to the ASCII) use a single octet with the same binary value as in ASCII." - http://en.wikipedia.org/wiki/UTF-8

[haru-s]

Your ideas are right at a point of binary semantics.
But some types, such as LPSTR of WAVEHDR.lpData, don't have character semantics.

For example;

import win32.mmsystem;
char* buffer = new char[BUFFER_LENGTH];
WAVEHDR header;
header.lpData = buffer;

You will have a error message "invalid UTF-8 sequence", because any strings are validated.

import win32.mmsystem;
ubyte[] buffer = new ubyte[BUFFER_LENGTH];
WAVEHDR header;
header.lpData = cast(LPSTR)buffer;

If a type of WAVEHDR.lpData is ubyte*, you will not use cast(LPSTR).

However, the cause of this problem is that Win32 API use char* to binary sequence, probably...