Index: meta.py
===================================================================
--- meta.py	(revision 321)
+++ meta.py	(working copy)
@@ -1268,12 +1268,22 @@
     order_by = ", ".join(order_by)
 
     # LIMIT and OFFSET clauses
-    if kwargs.get('limit') is not None:
-        limit_sql = " LIMIT %s " % kwargs['limit']
-        if kwargs.get('offset') is not None and kwargs['offset'] != 0:
-            limit_sql += "OFFSET %s " % kwargs['offset']
+    #gheorghe:
+    from django.conf.settings import DATABASE_ENGINE
+    limit_sql = ""
+    if DATABASE_ENGINE=="ado_mssql":
+        if kwargs.get('limit') is not None:
+            select[0] = 'TOP %s %s' % (kwargs['limit'], select[0])
+            if kwargs.get('offset') is not None and kwargs['offset'] != 0:
+                #the problem is if PK is not ID and also if user adds GROUP BY, HAVING etc (can a user do that?), 
+                #those should be added to the subquery too and they can't be 
+                #or at least I don't know how
+                where.append("id NOT IN (SELECT TOP %s id FROM %s%s)" % (kwargs['offset'], opts.db_table, (order_by and " ORDER BY " + order_by or "")))        
     else:
-        limit_sql = ""
+        if kwargs.get('limit') is not None:
+            limit_sql = " LIMIT %s " % kwargs['limit']
+            if kwargs.get('offset') is not None and kwargs['offset'] != 0:
+                limit_sql += "OFFSET %s " % kwargs['offset']
 
     return select, " FROM " + ",".join(tables) + (where and " WHERE " + " AND ".join(where) or "") + (order_by and " ORDER BY " + order_by or "") + limit_sql, params
 
@@ -1302,7 +1312,15 @@
     if field.null:
         kwargs.setdefault('where', []).append('%s.%s IS NOT NULL' % (opts.db_table, field.name))
     select, sql, params = function_get_sql_clause(opts, **kwargs)
-    sql = 'SELECT %s %s GROUP BY 1 ORDER BY 1' % (db.get_date_trunc_sql(kind, '%s.%s' % (opts.db_table, field.name)), sql)
+    
+    #gheorghe
+    from django.conf.settings import DATABASE_ENGINE
+    if DATABASE_ENGINE=="ado_mssql":
+        datepart = db.get_date_trunc_sql(kind, '%s.%s' % (opts.db_table, field.name))
+        sql = 'SELECT %s %s GROUP BY %s ORDER BY %s' % (datepart, sql, datepart, datepart)
+    else:
+        sql = 'SELECT %s %s GROUP BY 1 ORDER BY 1' % (db.get_date_trunc_sql(kind, '%s.%s' % (opts.db_table, field.name)), sql)
+    
     cursor = db.db.cursor()
     cursor.execute(sql, params)
     # We have to manually run typecast_timestamp(str()) on the results, because