diff --git a/include/wx/strconv.h b/include/wx/strconv.h
index 12346c9763..9aa87f8014 100644
--- a/include/wx/strconv.h
+++ b/include/wx/strconv.h
@@ -543,6 +543,39 @@ private:
     wxMBConv *m_convReal;
 };
 
+// ----------------------------------------------------------------------------
+// wxWhateverWorksConv: use whatever encoding works for the input
+// ----------------------------------------------------------------------------
+
+class WXDLLIMPEXP_BASE wxWhateverWorksConv : public wxMBConv
+{
+public:
+    wxWhateverWorksConv()
+    {
+    }
+
+    // Try to interpret the string as UTF-8, if it fails fall back to the
+    // current locale encoding (wxConvLibc) and if this fails as well,
+    // interpret it as wxConvISO8859_1 (which is used because it never fails
+    // and this conversion is used when we really, really must produce
+    // something on output).
+    virtual size_t
+    ToWChar(wchar_t *dst, size_t dstLen,
+            const char *src, size_t srcLen = wxNO_LEN) const wxOVERRIDE;
+
+    // Try to encode the string using the current locale encoding (wxConvLibc)
+    // and fall back to UTF-8 (which never fails) if it doesn't work. Note that
+    // we never use wxConvISO8859_1 here as we prefer to fall back on UTF-8
+    // even for the strings containing only code points representable in 8869-1.
+    virtual size_t
+    FromWChar(char *dst, size_t dstLen,
+              const wchar_t *src, size_t srcLen = wxNO_LEN) const wxOVERRIDE;
+
+    virtual wxMBConv *Clone() const wxOVERRIDE
+    {
+        return new wxWhateverWorksConv();
+    }
+};
 
 // ----------------------------------------------------------------------------
 // declare predefined conversion objects
@@ -578,6 +611,12 @@ WX_DECLARE_GLOBAL_CONV(wxMBConvStrictUTF8, wxConvUTF8)
 WX_DECLARE_GLOBAL_CONV(wxMBConvUTF7, wxConvUTF7)
 #define wxConvUTF7 wxGet_wxConvUTF7()
 
+// conversion used when we may not afford to lose data when outputting Unicode
+// strings (should be avoid in the other direction as it can misinterpret the
+// input encoding)
+WX_DECLARE_GLOBAL_CONV(wxWhateverWorksConv, wxConvWhateverWorks)
+#define wxConvWhateverWorks wxGet_wxConvWhateverWorks()
+
 // conversion used for the file names on the systems where they're not Unicode
 // (basically anything except Windows)
 //
diff --git a/interface/wx/strconv.h b/interface/wx/strconv.h
index 9388b49135..86e765d548 100644
--- a/interface/wx/strconv.h
+++ b/interface/wx/strconv.h
@@ -483,6 +483,30 @@ public:
     bool IsOk() const;
 };
 
+/**
+    Conversion object always producing non-empty output for non-empty input.
+
+    Conversions done using this object never lose data, at the cost of possibly
+    producing the output in an unwanted encoding or misinterpreting input
+    encoding.
+
+    To be precise, converting Unicode to multibyte strings using this object
+    tries to use the current locale encoding first but if this doesn't work, it
+    falls back to using UTF-8. In the other direction, UTF-8 is tried first,
+    then the current locale encoding and if this fails too, input is
+    interpreted as using ISO 8859-1, which never fails.
+
+    It is almost always @e wrong to use this converter for multibyte-to-Unicode
+    direction as the program should know which encoding the input data is
+    supposed to use and use the appropriate converter instead. However it may
+    be useful in the Unicode-to-multibyte direction if the goal is to produce
+    the output in the current locale encoding if possible, but still output
+    something, instead of nothing at all, even if the Unicode string is not
+    representable in this encoding.
+
+    @since 3.1.0
+ */
+extern wxMBConv& wxConvWhateverWorks;
 
 
 /**
diff --git a/src/common/strconv.cpp b/src/common/strconv.cpp
index 9e7da47925..2756870d55 100644
--- a/src/common/strconv.cpp
+++ b/src/common/strconv.cpp
@@ -3286,6 +3286,40 @@ bool wxCSConv::IsUTF8() const
 #endif
 
 
+// ============================================================================
+// wxWhateverWorksConv
+// ============================================================================
+
+size_t
+wxWhateverWorksConv::ToWChar(wchar_t *dst, size_t dstLen,
+                             const char *src, size_t srcLen) const
+{
+    size_t rc = wxConvUTF8.ToWChar(dst, dstLen, src, srcLen);
+    if ( rc != wxCONV_FAILED )
+        return rc;
+
+    rc = wxConvLibc.ToWChar(dst, dstLen, src, srcLen);
+    if ( rc != wxCONV_FAILED )
+        return rc;
+
+    rc = wxConvISO8859_1.ToWChar(dst, dstLen, src, srcLen);
+
+    return rc;
+}
+
+size_t
+wxWhateverWorksConv::FromWChar(char *dst, size_t dstLen,
+                               const wchar_t *src, size_t srcLen) const
+{
+    size_t rc = wxConvLibc.FromWChar(dst, dstLen, src, srcLen);
+    if ( rc != wxCONV_FAILED )
+        return rc;
+
+    rc = wxConvUTF8.FromWChar(dst, dstLen, src, srcLen);
+
+    return rc;
+}
+
 #if wxUSE_UNICODE
 
 wxWCharBuffer wxSafeConvertMB2WX(const char *s)
@@ -3330,6 +3364,7 @@ wxCharBuffer wxSafeConvertWX2MB(const wchar_t *ws)
 #undef wxConvLibc
 #undef wxConvUTF8
 #undef wxConvUTF7
+#undef wxConvWhateverWorks
 #undef wxConvLocal
 #undef wxConvISO8859_1
 
@@ -3369,6 +3404,7 @@ wxCharBuffer wxSafeConvertWX2MB(const wchar_t *ws)
 //     empty statement (and hope that no compilers warns about this)
 WX_DEFINE_GLOBAL_CONV(wxMBConvStrictUTF8, wxConvUTF8, ;);
 WX_DEFINE_GLOBAL_CONV(wxMBConvUTF7, wxConvUTF7, ;);
+WX_DEFINE_GLOBAL_CONV(wxWhateverWorksConv, wxConvWhateverWorks, ;);
 
 WX_DEFINE_GLOBAL_CONV(wxCSConv, wxConvLocal, (wxFONTENCODING_SYSTEM));
 WX_DEFINE_GLOBAL_CONV(wxCSConv, wxConvISO8859_1, (wxFONTENCODING_ISO8859_1));
@@ -3387,5 +3423,5 @@ WXDLLIMPEXP_DATA_BASE(wxMBConv *) wxConvFileName =
 #ifdef __DARWIN__
                                     &wxConvMacUTF8DObj;
 #else // !__DARWIN__
-                                    wxGet_wxConvLibcPtr();
+                                    wxGet_wxConvWhateverWorksPtr();
 #endif // __DARWIN__/!__DARWIN__