fixed decodeUTF8 for pgf; removed old resources from darcs

lib/resource/interlingua/RelativeIna.gf

@@ -0,0 +1,53 @@
+concrete RelativeIna of Relative = CatIna ** open ResIna in {
+
+  flags optimize=all_subs ;
+
+  lin
+
+    RelCl cl = {
+      s = \\t,a,p,agr => 
+	(case agr.n of {Sg => "tal"; Pl => "tales"}) ++ 
+	"que" ++ cl.s ! t ! a ! p ! ODir ; 
+      c = Nom
+      } ;
+
+    RelVP rp vp = {
+      s = \\t,a,p,agr => (mkClause (rp.s!Nom) agr vp).s ! t ! a ! p ! ODir;
+      c = Nom
+      } ; 
+    -- !!! person agreement is probably bad here; see below.
+
+---- Pied piping: "a que tu invia flores"
+
+    RelSlash rp slash = {
+      s = \\t,a,p,agr => slash.p2 ++ rp.s ! slash.c2 ++ slash.s ! t ! a ! p ! ODir ;
+      c = slash.c2; 
+      } ; 
+    -- !!! In the above The agreement feature of the RP does not match
+    -- the the (parametric!) agreement of the resulting clause.  
+    -- Is it a bug? I believe there is the same behaviour in the english grammar.
+
+    FunRP p np rp = {
+      s = \\c => np.s ! Acc ++ p.s ++ rp.s ! p.c ;
+      a = np.a
+      } ;
+
+
+    IdRP = {
+	-- TODO: variant: "le qual"
+      a = {p = P3; n = variants {Sg; Pl}};
+      s = table {
+	Nom => quique;   --  Le ultime traino que pote portar me ibi a tempore parti in cinque minutas
+	Gen => "cuje";  --  Le documentos cuje importantia esseva dubitose incriminava le spia
+	Acc => "que";   --  Le documentos que le spia portava con se esseva multo importante 
+	Dat => "a" ++ quique;
+	Abl => "de" ++ quique
+      }} ;
+
+    oper 
+      quique = variants {"qui";  -- !!! Only for humans, only after a preposition.
+			 -- This is extremely strange, because it does not match any Romance language I know. 
+			 -- For now just be lax and make it a variant of "que"
+			 "que"};
+      
+}