w32: Always use UTF-8 for localized texts

* configure.ac: Define GPG_ERR_ENABLE_GETTEXT_MACROS.
* src/version.c (do_subsystem_inits) [W32]: Switch gettext to UTF-8.
--

On Windows, applications using GpgME can now rely on localized texts
returned by GpgME, e.g. the texts for error codes, to be UTF-8 encoded.
In particular, this saves a useless and possibly lossy conversion to
native encoding and back to UTF-8. On other OSes, where we use the
system provided gettext, we assume that UTF-8 is used nowadays.

GnuPG-bug-id: 5960

NEWS

@@ -1,6 +1,9 @@
 Noteworthy changes in version 1.20.0 (unreleased)
 -------------------------------------------------
 
+ * On Windows, the gettext functions provided by gpgrt are switched into utf8
+   mode, so that all localized texts returned by GpgME or gpgrt, e.g. the
+   texts for error codes are now UTF-8 encoded.
 
  * Interface changes relative to the 1.19.0 release:
  ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

configure.ac

@@ -922,6 +922,9 @@ AH_BOTTOM([
 # define GPGME_GCC_A_PURE
 #endif
 
+/* Under Windows we use the gettext code from gpgrt.  */
+#define GPG_ERR_ENABLE_GETTEXT_MACROS
+
 /* Under WindowsCE we need gpg-error's strerror macro.  */
 #define GPG_ERR_ENABLE_ERRNO_MACROS 1
 

src/version.c

@@ -72,6 +72,9 @@ do_subsystem_inits (void)
 
     WSAStartup (0x202, &wsadat);
   }
+
+  /* We want gpgrt's gettext to always output UTF-8. */
+  gettext_use_utf8 (1);
 #endif
 
   _gpgme_debug_subsystem_init ();