Use `signed char' explicitly where needed. It is important because:

- It is implementation-dependent if plain `char' signed or not.
- C standard defines three *distinct* types: char, signed char,
  and unsigned char.
- Assuming that char is always unsigned or signed can lead to
  compile-time and run-time errors.

You can also use int8_t, but then it would be less obvious for developers
to never assume that char is always unsigned (or always signed).

Conflicts:

	components/esm/loadcell.hpp

apps/esmtool/labels.cpp

@@ -16,7 +16,7 @@
 #include <iostream>
 #include <boost/format.hpp>
 
-std::string bodyPartLabel(char idx)
+std::string bodyPartLabel(signed char idx)
 {
     const char *bodyPartLabels[] =  {
         "Head",
@@ -47,14 +47,14 @@ std::string bodyPartLabel(char idx)
         "Weapon",
         "Tail"
     };
-    
+
     if ((int)idx >= 0 && (int)(idx) <= 26)
         return bodyPartLabels[(int)(idx)];
     else
         return "Invalid";
 }
 
-std::string meshPartLabel(char idx)
+std::string meshPartLabel(signed char idx)
 {
     const char *meshPartLabels[] =  {
         "Head",
@@ -73,25 +73,25 @@ std::string meshPartLabel(char idx)
         "Clavicle",
         "Tail"
     };
-    
+
     if ((int)(idx) >= 0 && (int)(idx) <= ESM::BodyPart::MP_Tail)
         return meshPartLabels[(int)(idx)];
     else
         return "Invalid";
 }
 
-std::string meshTypeLabel(char idx)
+std::string meshTypeLabel(signed char idx)
 {
     const char *meshTypeLabels[] =  {
         "Skin",
         "Clothing",
         "Armor"
     };
-    
+
     if ((int)(idx) >= 0 && (int)(idx) <= ESM::BodyPart::MT_Armor)
         return meshTypeLabels[(int)(idx)];
     else
-        return "Invalid";   
+        return "Invalid";
 }
 
 std::string clothingTypeLabel(int idx)
@@ -108,7 +108,7 @@ std::string clothingTypeLabel(int idx)
         "Ring",
         "Amulet"
     };
-    
+
     if (idx >= 0 && idx <= 9)
         return clothingTypeLabels[idx];
     else

apps/esmtool/labels.hpp

@@ -3,9 +3,9 @@
 
 #include <string>
 
-std::string bodyPartLabel(char idx);
-std::string meshPartLabel(char idx);
-std::string meshTypeLabel(char idx);
+std::string bodyPartLabel(signed char idx);
+std::string meshPartLabel(signed char idx);
+std::string meshTypeLabel(signed char idx);
 std::string clothingTypeLabel(int idx);
 std::string armorTypeLabel(int idx);
 std::string dialogTypeLabel(int idx);

components/esm/loadcell.hpp

@@ -77,8 +77,8 @@ public:
   std::string mKey, mTrap; // Key and trap ID names, if any
 
   // No idea - occurs ONCE in Morrowind.esm, for an activator
-  char mUnam;
-  
+  signed char mUnam;
+
   // Track deleted references. 0 - not deleted, 1 - deleted, but respawns, 2 - deleted and does not respawn.
   int mDeleted;
 
@@ -164,14 +164,14 @@ struct Cell
   bool mWaterInt;
   int mMapColor;
   int mNAM0;
-  
+
   // References "leased" from another cell (i.e. a different cell
   //  introduced this ref, and it has been moved here by a plugin)
   CellRefTracker mLeasedRefs;
   MovedCellRefTracker mMovedRefs;
-  
+
   void load(ESMReader &esm, MWWorld::ESMStore &store);
-  
+
   // This method is left in for compatibility with esmtool. Parsing moved references currently requires
   //  passing ESMStore, bit it does not know about this parameter, so we do it this way.
   void load(ESMReader &esm) {};
@@ -209,7 +209,7 @@ struct Cell
      reuse one memory location without blanking it between calls.
   */
   static bool getNextRef(ESMReader &esm, CellRef &ref);
-  
+
   /* This fetches an MVRF record, which is used to track moved references.
    * Since they are comparably rare, we use a separate method for this.
    */

components/esm/loaddial.hpp

@@ -30,7 +30,7 @@ struct Dialogue
     };
 
     std::string mId;
-    char mType;
+    signed char mType;
     std::vector<DialInfo> mInfo;
 
     void load(ESMReader &esm);

components/to_utf8/tables_gen.hpp

@@ -8,7 +8,7 @@ namespace ToUTF8
 /// Central European and Eastern European languages that use Latin script,
 /// such as Polish, Czech, Slovak, Hungarian, Slovene, Bosnian, Croatian,
 /// Serbian (Latin script), Romanian and Albanian.
-static char windows_1250[] =
+static signed char windows_1250[] =
 {
     1, 0, 0, 0, 0, 0,
     1, 1, 0, 0, 0, 0,
@@ -270,7 +270,7 @@ static char windows_1250[] =
 
 /// Cyrillic alphabet such as Russian, Bulgarian, Serbian Cyrillic
 /// and other languages
-static char windows_1251[] =
+static signed char windows_1251[] =
 {
     1, 0, 0, 0, 0, 0,
     1, 1, 0, 0, 0, 0,
@@ -531,7 +531,7 @@ static char windows_1251[] =
 };
 
 /// Latin alphabet used by English and some other Western languages
-static char windows_1252[] =
+static signed char windows_1252[] =
 {
     1, 0, 0, 0, 0, 0,
     1, 1, 0, 0, 0, 0,

components/to_utf8/to_utf8.cpp

@@ -216,7 +216,7 @@ void Utf8Encoder::copyFromArray(unsigned char ch, char* &out)
         return;
     }
 
-    const char *in = translationArray + ch*6;
+    const signed char *in = translationArray + ch*6;
     int len = *(in++);
     for (int i=0; i<len; i++)
         *(out++) = *(in++);

components/to_utf8/to_utf8.hpp

@@ -47,7 +47,7 @@ namespace ToUTF8
             void copyFromArray2(const char*& chp, char* &out);
 
             std::vector<char> mOutput;
-            char* translationArray;
+            signed char* translationArray;
     };
 }