I am trying to find a substitute for LPWSTR for porting a project to gcc.
typedef __nullterminated WCHAR *NWPSTR, *LPWSTR, *PWSTR;
What is null terminated ? so would it be safe if I did something like this:
typedef WCHAR *LPWSTR
typedef __nullterminated WCHAR *NWPSTR, *LPWSTR, *PWSTR;
The __nullterminated
part is a SAL annotation. SAL is a Microsoft specific technology to annotate function parameters, return values, function behaviors, etc. to help finding bugs and reduce C/C++ code defects using the Visual Studio Code Analysis tool. You can read about SAL here on MSDN:
WCHAR
is defined in the Windows Platform SDK headers basically as a typedef for wchar_t
, which is a 16-bit character type in the Microsoft Visual C++ compiler. This is used as a "character unit" for the Unicode UTF-16 encoding, which is the default de facto Unicode encoding of Windows APIs.
Note that other compilers like GNU GCC on Linux consider wchar_t
to be a 32-bit character unit (not 16-bit), so you have to pay attention here for the portability of your code.
(Note: Modern versions of Windows support Unicode UTF-16 with the so called surrogate pairs, so you can have a couple of adjacent WCHAR
s defining a surrogate pair.)
NWPSTR
, LPWSTR
and PWSTR
are all synonyms, defined as pointers to WCHAR
, i.e. considering also the __nullterminated
SAL annotation, they are pointers to "raw" C-style NUL-terminated Unicode UTF-16 strings.
Basically, this is the Windows Win32 Unicode equivalent of the classical C's char*
.
I've been programming Windows in C++ for several years, and I've never met this NWPSTR
to be honest :)
The name PWSTR
is built using the following elements:
P
: pointerW
: "wide", i.e. WCHAR
/wchar_t
-based, i.e. Unicode UTF-16STR
: stringSo, PWSTR
means a pointer to a WCHAR
/wchar_t
string, i.e. a Unicode UTF-16 string, as already stated above.
LPWSTR
is just an old name for the same thing; the initial L
means "long", and that dates back to a time when there were "long pointers" which could access memory farther than "short" or "near" pointers :) Those days are no more.
And, if you put a C
in those names, you have typedefs for the read-only const
counterparts, e.g.: PCWSTR
or LPCWSTR
are basically const wchar_t*
NUL-terminated Unicode UTF-16 strings.
You will find PCWSTR
and the older equivalent LPCWSTR
used a lot in Windows header files and Windows API documentation to represent Unicode UTF-16 "input" (i.e. const
) C-style NUL-terminated strings.