Top |
Functions
Description
These functions provide some level of UNIX emulation on the Windows platform. If your application really needs the POSIX APIs, we suggest you try the Cygwin project.
Functions
g_win32_get_command_line ()
gchar **
g_win32_get_command_line (void
);
Gets the command line arguments, on Windows, in the GLib filename encoding (ie: UTF-8).
Normally, on Windows, the command line arguments are passed to main()
in the system codepage encoding. This prevents passing filenames as
arguments if the filenames contain characters that fall outside of
this codepage. If such filenames are passed, then substitutions
will occur (such as replacing some characters with '?').
GLib's policy of using UTF-8 as a filename encoding on Windows was
designed to localise the pain of dealing with filenames outside of
the system codepage to one area: dealing with commandline arguments
in main()
.
As such, most GLib programs should ignore the value of argv passed to
their main()
function and call g_win32_get_command_line()
instead.
This will get the "full Unicode" commandline arguments using
GetCommandLineW()
and convert it to the GLib filename encoding (which
is UTF-8 on Windows).
The strings returned by this function are suitable for use with
functions such as g_open()
and g_file_new_for_commandline_arg()
but
are not suitable for use with g_option_context_parse()
, which assumes
that its input will be in the system codepage. The return value is
suitable for use with g_option_context_parse_strv()
, however, which
is a better match anyway because it won't leak memory.
Unlike argv, the returned value is a normal strv and can (and should)
be freed with g_strfreev()
when no longer needed.
Since 2.40
g_win32_error_message ()
gchar *
g_win32_error_message (gint error
);
Translate a Win32 error code (as returned by GetLastError()
or
WSAGetLastError()
) into the corresponding message. The message is
either language neutral, or in the thread's language, or the user's
language, the system's language, or US English (see docs for
FormatMessage()
). The returned string is in UTF-8. It should be
deallocated with g_free()
.
g_win32_getlocale ()
gchar *
g_win32_getlocale (void
);
The setlocale()
function in the Microsoft C library uses locale
names of the form "English_United States.1252" etc. We want the
UNIXish standard form "en_US", "zh_TW" etc. This function gets the
current thread locale from Windows - without any encoding info -
and returns it as a string of the above form for use in forming
file names etc. The returned string should be deallocated with
g_free()
.
g_win32_get_package_installation_directory ()
gchar * g_win32_get_package_installation_directory (const gchar *package
,const gchar *dll_name
);
g_win32_get_package_installation_directory
has been deprecated since version 2.18 and should not be used in newly-written code.
Pass the HMODULE of a DLL or EXE to
g_win32_get_package_installation_directory_of_module()
instead.
Try to determine the installation directory for a software package.
This function is deprecated. Use
g_win32_get_package_installation_directory_of_module()
instead.
The use of package
is deprecated. You should always pass NULL
. A
warning is printed if non-NULL is passed as package
.
The original intended use of package
was for a short identifier of
the package, typically the same identifier as used for
GETTEXT_PACKAGE
in software configured using GNU
autotools. The function first looks in the Windows Registry for the
value #InstallationDirectory
in the key
#HKLM\Software\@package
, and if that value
exists and is a string, returns that.
It is strongly recommended that packagers of GLib-using libraries for Windows do not store installation paths in the Registry to be used by this function as that interfers with having several parallel installations of the library. Enabling multiple installations of different versions of some GLib-using library, or GLib itself, is desirable for various reasons.
For this reason it is recommeded to always pass NULL
as
package
to this function, to avoid the temptation to use the
Registry. In version 2.20 of GLib the package
parameter
will be ignored and this function won't look in the Registry at all.
If package
is NULL
, or the above value isn't found in the
Registry, but dll_name
is non-NULL
, it should name a DLL loaded
into the current process. Typically that would be the name of the
DLL calling this function, looking for its installation
directory. The function then asks Windows what directory that DLL
was loaded from. If that directory's last component is "bin" or
"lib", the parent directory is returned, otherwise the directory
itself. If that DLL isn't loaded, the function proceeds as if
dll_name
was NULL
.
If both package
and dll_name
are NULL
, the directory from where
the main executable of the process was loaded is used instead in
the same way as above.
g_win32_get_package_installation_directory_of_module ()
gchar *
g_win32_get_package_installation_directory_of_module
(gpointer hmodule
);
This function tries to determine the installation directory of a software package based on the location of a DLL of the software package.
hmodule
should be the handle of a loaded DLL or NULL
. The
function looks up the directory that DLL was loaded from. If
hmodule
is NULL, the directory the main executable of the current
process is looked up. If that directory's last component is "bin"
or "lib", its parent directory is returned, otherwise the directory
itself.
It thus makes sense to pass only the handle to a "public" DLL of a software package to this function, as such DLLs typically are known to be installed in a "bin" or occasionally "lib" subfolder of the installation folder. DLLs that are of the dynamically loaded module or plugin variety are often located in more private locations deeper down in the tree, from which it is impossible for GLib to deduce the root of the package installation.
The typical use case for this function is to have a DllMain()
that
saves the handle for the DLL. Then when code in the DLL needs to
construct names of files in the installation tree it calls this
function passing the DLL handle.
Parameters
hmodule |
The Win32 handle for a DLL loaded into the current process, or |
[allow-none] |
Returns
a string containing the guessed installation directory for
the software package hmodule
is from. The string is in the GLib
file name encoding, i.e. UTF-8. The return value should be freed
with g_free()
when not needed any longer. If the function fails
NULL
is returned.
Since 2.16
g_win32_get_package_installation_subdirectory ()
gchar * g_win32_get_package_installation_subdirectory (const gchar *package
,const gchar *dll_name
,const gchar *subdir
);
g_win32_get_package_installation_subdirectory
has been deprecated since version 2.18 and should not be used in newly-written code.
Pass the HMODULE of a DLL or EXE to
g_win32_get_package_installation_directory_of_module()
instead, and
then construct a subdirectory pathname with g_build_filename()
.
This function is deprecated. Use
g_win32_get_package_installation_directory_of_module()
and
g_build_filename()
instead.
Returns a newly-allocated string containing the path of the
subdirectory subdir
in the return value from calling
g_win32_get_package_installation_directory()
with the package
and
dll_name
parameters. See the documentation for
g_win32_get_package_installation_directory()
for more details. In
particular, note that it is deprecated to pass anything except NULL
as package
.
g_win32_get_windows_version ()
guint
g_win32_get_windows_version (void
);
Returns version information for the Windows operating system the
code is running on. See MSDN documentation for the GetVersion()
function. To summarize, the most significant bit is one on Win9x,
and zero on NT-based systems. Since version 2.14, GLib works only
on NT-based systems, so checking whether your are running on Win9x
in your own software is moot. The least significant byte is 4 on
Windows NT 4, and 5 on Windows XP. Software that needs really
detailed version and feature information should use Win32 API like
GetVersionEx()
and VerifyVersionInfo()
.
Since 2.6
g_win32_locale_filename_from_utf8 ()
gchar *
g_win32_locale_filename_from_utf8 (const gchar *utf8filename
);
Converts a filename from UTF-8 to the system codepage.
On NT-based Windows, on NTFS file systems, file names are in Unicode. It is quite possible that Unicode file names contain characters not representable in the system codepage. (For instance, Greek or Cyrillic characters on Western European or US Windows installations, or various less common CJK characters on CJK Windows installations.)
In such a case, and if the filename refers to an existing file, and
the file system stores alternate short (8.3) names for directory
entries, the short form of the filename is returned. Note that the
"short" name might in fact be longer than the Unicode name if the
Unicode name has very short pathname components containing
non-ASCII characters. If no system codepage name for the file is
possible, NULL
is returned.
The return value is dynamically allocated and should be freed with
g_free()
when no longer needed.
Since 2.8
G_WIN32_DLLMAIN_FOR_DLL_NAME()
# define G_WIN32_DLLMAIN_FOR_DLL_NAME(static, dll_name)
G_WIN32_DLLMAIN_FOR_DLL_NAME
is deprecated and should not be used in newly-written code.
On Windows, this macro defines a DllMain()
function that stores
the actual DLL name that the code being compiled will be included in.
On non-Windows platforms, expands to nothing.
G_WIN32_HAVE_WIDECHAR_API
#define G_WIN32_HAVE_WIDECHAR_API() TRUE
On Windows, this macro defines an expression which evaluates to
TRUE
if the code is running on a version of Windows where the wide
character versions of the Win32 API functions, and the wide character
versions of the C library functions work. (They are always present in
the DLLs, but don't work on Windows 9x and Me.)
On non-Windows platforms, it is not defined.
Since 2.6
G_WIN32_IS_NT_BASED
#define G_WIN32_IS_NT_BASED() TRUE
On Windows, this macro defines an expression which evaluates to
TRUE
if the code is running on an NT-based Windows operating system.
On non-Windows platforms, it is not defined.
Since 2.6
Types and Values
MAXPATHLEN
#define MAXPATHLEN 1024
Provided for UNIX emulation on Windows; equivalent to UNIX
macro MAXPATHLEN
, which is the maximum length of a filename
(including full path).