Always allow writing to _ (dontcare)

`_` is now an identifier that is always valid to write to, but never to
read from. It can also be shadowed freely.

Closes #876.

pony.g

@@ -135,13 +135,12 @@ elseif
 
 idseq
   : ID
-  | '_'
-  | ('(' | LPAREN_NEW) (idseq_in_seq | '_') (',' (idseq_in_seq | '_'))* ')'
+  | ('(' | LPAREN_NEW) idseq_in_seq (',' idseq_in_seq)* ')'
   ;
 
 idseq_in_seq
   : ID
-  | ('(' | LPAREN_NEW) (idseq_in_seq | '_') (',' (idseq_in_seq | '_'))* ')'
+  | ('(' | LPAREN_NEW) idseq_in_seq (',' idseq_in_seq)* ')'
   ;
 
 nextpattern
@@ -192,7 +191,7 @@ nextatom
   : ID
   | 'this'
   | literal
-  | LPAREN_NEW (rawseq | '_') tuple? ')'
+  | LPAREN_NEW rawseq tuple? ')'
   | LSQUARE_NEW ('as' type ':')? rawseq (',' rawseq)* ']'
   | 'object' ('\' ID (',' ID)* '\')? cap? ('is' type)? members 'end'
   | '{' ('\' ID (',' ID)* '\')? cap? ID? typeparams? ('(' | LPAREN_NEW) params? ')' lambdacaptures? (':' type)? '?'? '=>' rawseq '}' cap?
@@ -205,7 +204,7 @@ atom
   : ID
   | 'this'
   | literal
-  | ('(' | LPAREN_NEW) (rawseq | '_') tuple? ')'
+  | ('(' | LPAREN_NEW) rawseq tuple? ')'
   | ('[' | LSQUARE_NEW) ('as' type ':')? rawseq (',' rawseq)* ']'
   | 'object' ('\' ID (',' ID)* '\')? cap? ('is' type)? members 'end'
   | '{' ('\' ID (',' ID)* '\')? cap? ID? typeparams? ('(' | LPAREN_NEW) params? ')' lambdacaptures? (':' type)? '?'? '=>' rawseq '}' cap?
@@ -215,7 +214,7 @@ atom
   ;
 
 tuple
-  : ',' (rawseq | '_') (',' (rawseq | '_'))*
+  : ',' rawseq (',' rawseq)*
   ;
 
 lambdacaptures
@@ -245,7 +244,7 @@ type
 atomtype
   : 'this'
   | cap
-  | ('(' | LPAREN_NEW) (infixtype | '_') tupletype? ')'
+  | ('(' | LPAREN_NEW) infixtype tupletype? ')'
   | nominal
   | lambdatype
   ;
@@ -255,7 +254,7 @@ lambdatype
   ;
 
 tupletype
-  : ',' (infixtype | '_') (',' (infixtype | '_'))*
+  : ',' infixtype (',' infixtype)*
   ;
 
 infixtype
@@ -322,7 +321,7 @@ literal
   ;
 
 param
-  : (parampattern | '_') (':' type)? ('=' infix)?
+  : parampattern (':' type)? ('=' infix)?
   ;
 
 antlr_0
@@ -379,7 +378,7 @@ Type:
 
 ID
   : LETTER (LETTER | DIGIT | '_' | '\'')*
-  | '_' (LETTER | DIGIT | '_' | '\'')+
+  | '_' (LETTER | DIGIT | '_' | '\'')*
   ;
 
 INT

src/libponyc/ast/ast.c

@@ -1438,7 +1438,7 @@ static void print_type(printbuf_t* buffer, ast_t* type)
       printbuf(buffer, "this");
       break;
 
-    case TK_DONTCARE:
+    case TK_DONTCARETYPE:
       printbuf(buffer, "_");
       break;
 

src/libponyc/ast/bnfprint.c

@@ -61,7 +61,7 @@ static const char* antlr_post =
   "// Lexer\n\n"
   "ID\n"
   "  : LETTER (LETTER | DIGIT | '_' | '\\'')*\n"
-  "  | '_' (LETTER | DIGIT | '_' | '\\'')+\n"
+  "  | '_' (LETTER | DIGIT | '_' | '\\'')*\n"
   "  ;\n\n"
   "INT\n"
   "  : DIGIT (DIGIT | '_')*\n"

src/libponyc/ast/id.c

@@ -27,6 +27,17 @@ bool check_id(pass_opt_t* opt, ast_t* id_node, const char* desc, int spec)
     name++;
     prev = '_';
 
+    if(*name == '\0')
+    {
+      if((spec & ALLOW_DONTCARE) == 0)
+      {
+        ast_error(opt->check.errors, id_node,
+          "%s name cannot be \"%s\"", desc, ast_name(id_node));
+        return false;
+      }
+      return true;
+    }
+
     if((spec & ALLOW_LEADING_UNDERSCORE) == 0)
     {
       ast_error(opt->check.errors, id_node,
@@ -168,9 +179,9 @@ bool check_id_param(pass_opt_t* opt, ast_t* id_node)
 
 bool check_id_local(pass_opt_t* opt, ast_t* id_node)
 {
-  // [a-z][A-Za-z0-9_]*'* (and no double or trailing underscores)
+  // (_|[a-z][A-Za-z0-9_]*'*) (and no double or trailing underscores)
   return check_id(opt, id_node, "local variable",
-    START_LOWER | ALLOW_UNDERSCORE | ALLOW_TICK);
+    START_LOWER | ALLOW_UNDERSCORE | ALLOW_TICK | ALLOW_DONTCARE);
 }
 
 bool is_name_type(const char* name)
@@ -186,7 +197,8 @@ bool is_name_type(const char* name)
 
 bool is_name_private(const char* name)
 {
-  return name[0] == '_' || (is_name_internal_test(name) && name[1] == '_');
+  return ((name[0] == '_') && (name[1] != '\0')) ||
+    (is_name_internal_test(name) && (name[1] == '_'));
 }
 
 bool is_name_ffi(const char* name)
@@ -198,3 +210,8 @@ bool is_name_internal_test(const char* name)
 {
   return name[0] == '$';
 }
+
+bool is_name_dontcare(const char* name)
+{
+  return (name[0] == '_') && (name[1] == '\0');
+}

src/libponyc/ast/id.h

@@ -47,6 +47,9 @@ bool is_name_ffi(const char* name);
 // Report whether the given id name is for an internal test
 bool is_name_internal_test(const char* name);
 
+// Report whether the given id name is '_'
+bool is_name_dontcare(const char* name);
+
 PONY_EXTERN_C_END
 
 #endif

src/libponyc/ast/id_internal.h

@@ -15,6 +15,7 @@ PONY_EXTERN_C_BEGIN
 #define ALLOW_LEADING_UNDERSCORE  0x04
 #define ALLOW_UNDERSCORE          0x08
 #define ALLOW_TICK                0x10
+#define ALLOW_DONTCARE            0x20
 
 
 /* Check that the name in the given ID node meets the given spec.

src/libponyc/ast/lexer.c

@@ -106,7 +106,6 @@ static const lextoken_t symbols[] =
 
 static const lextoken_t keywords[] =
 {
-  { "_", TK_DONTCARE },
   { "compile_intrinsic", TK_COMPILE_INTRINSIC },
 
   { "use", TK_USE },
@@ -211,6 +210,7 @@ static const lextoken_t abstract[] =
   { "members", TK_MEMBERS },
   { "fvar", TK_FVAR },
   { "flet", TK_FLET },
+  { "dontcare", TK_DONTCARE },
   { "ffidecl", TK_FFIDECL },
   { "fficall", TK_FFICALL },
 
@@ -222,6 +222,7 @@ static const lextoken_t abstract[] =
   { "thistype", TK_THISTYPE },
   { "funtype", TK_FUNTYPE },
   { "lambdatype", TK_LAMBDATYPE },
+  { "dontcaretype", TK_DONTCARETYPE },
   { "infer", TK_INFERTYPE },
   { "errortype", TK_ERRORTYPE },
 
@@ -279,6 +280,7 @@ static const lextoken_t abstract[] =
   { "varref", TK_VARREF },
   { "letref", TK_LETREF },
   { "paramref", TK_PARAMREF },
+  { "dontcareref", TK_DONTCAREREF },
   { "newapp", TK_NEWAPP },
   { "beapp", TK_BEAPP },
   { "funapp", TK_FUNAPP },

src/libponyc/ast/parser.c

@@ -57,23 +57,17 @@ DECL(thisliteral);
 
 // Parse rules
 
-// DONTCARE
-DEF(dontcare);
-  PRINT_INLINE();
-  TOKEN(NULL, TK_DONTCARE);
-  DONE();
-
 // type
 DEF(provides);
   PRINT_INLINE();
   AST_NODE(TK_PROVIDES);
   RULE("provided type", type);
   DONE();
 
-// (postfix | dontcare) [COLON type] [ASSIGN infix]
+// postfix [COLON type] [ASSIGN infix]
 DEF(param);
   AST_NODE(TK_PARAM);
-  RULE("name", parampattern, dontcare);
+  RULE("name", parampattern);
   IF(TK_COLON,
     RULE("parameter type", type);
     UNWRAP(0, TK_REFERENCE);
@@ -196,20 +190,20 @@ DEF(infixtype);
   SEQ("type", uniontype, isecttype);
   DONE();
 
-// COMMA (infixtype | dontcare) {COMMA (infixtype | dontcare)}
+// COMMA infixtype {COMMA infixtype}
 DEF(tupletype);
   INFIX_BUILD();
   TOKEN(NULL, TK_COMMA);
   MAP_ID(TK_COMMA, TK_TUPLETYPE);
-  RULE("type", infixtype, dontcare);
-  WHILE(TK_COMMA, RULE("type", infixtype, dontcare));
+  RULE("type", infixtype);
+  WHILE(TK_COMMA, RULE("type", infixtype));
   DONE();
 
-// (LPAREN | LPAREN_NEW) (infixtype | dontcare) [tupletype] RPAREN
+// (LPAREN | LPAREN_NEW) infixtype [tupletype] RPAREN
 DEF(groupedtype);
   PRINT_INLINE();
   SKIP(NULL, TK_LPAREN, TK_LPAREN_NEW);
-  RULE("type", infixtype, dontcare);
+  RULE("type", infixtype);
   OPT_NO_DFLT RULE("type", tupletype);
   SKIP(NULL, TK_RPAREN);
   SET_FLAG(AST_FLAG_IN_PARENS);
@@ -413,30 +407,30 @@ DEF(nextarray);
   TERMINATE("array literal", TK_RSQUARE);
   DONE();
 
-// COMMA (rawseq | dontcare) {COMMA (rawseq | dontcare)}
+// COMMA rawseq {COMMA rawseq}
 DEF(tuple);
   INFIX_BUILD();
   TOKEN(NULL, TK_COMMA);
   MAP_ID(TK_COMMA, TK_TUPLE);
-  RULE("value", rawseq, dontcare);
-  WHILE(TK_COMMA, RULE("value", rawseq, dontcare));
+  RULE("value", rawseq);
+  WHILE(TK_COMMA, RULE("value", rawseq));
   DONE();
 
-// (LPAREN | LPAREN_NEW) (rawseq | dontcare) [tuple] RPAREN
+// (LPAREN | LPAREN_NEW) rawseq [tuple] RPAREN
 DEF(groupedexpr);
   PRINT_INLINE();
   SKIP(NULL, TK_LPAREN, TK_LPAREN_NEW);
-  RULE("value", rawseq, dontcare);
+  RULE("value", rawseq);
   OPT_NO_DFLT RULE("value", tuple);
   SKIP(NULL, TK_RPAREN);
   SET_FLAG(AST_FLAG_IN_PARENS);
   DONE();
 
-// LPAREN_NEW (rawseq | dontcare) [tuple] RPAREN
+// LPAREN_NEW rawseq [tuple] RPAREN
 DEF(nextgroupedexpr);
   PRINT_INLINE();
   SKIP(NULL, TK_LPAREN_NEW);
-  RULE("value", rawseq, dontcare);
+  RULE("value", rawseq);
   OPT_NO_DFLT RULE("value", tuple);
   SKIP(NULL, TK_RPAREN);
   SET_FLAG(AST_FLAG_IN_PARENS);
@@ -587,8 +581,8 @@ DEF(idseqmulti);
   PRINT_INLINE();
   AST_NODE(TK_TUPLE);
   SKIP(NULL, TK_LPAREN, TK_LPAREN_NEW);
-  RULE("variable name", idseq_in_seq, dontcare);
-  WHILE(TK_COMMA, RULE("variable name", idseq_in_seq, dontcare));
+  RULE("variable name", idseq_in_seq);
+  WHILE(TK_COMMA, RULE("variable name", idseq_in_seq));
   SKIP(NULL, TK_RPAREN);
   DONE();
 
@@ -606,9 +600,9 @@ DEF(idseq_in_seq);
   RULE("variable name", idseqsingle, idseqmulti);
   DONE();
 
-// ID | '_' | (LPAREN | LPAREN_NEW) idseq {COMMA idseq} RPAREN
+// ID | (LPAREN | LPAREN_NEW) idseq {COMMA idseq} RPAREN
 DEF(idseq);
-  RULE("variable name", idseqsingle, dontcare, idseqmulti);
+  RULE("variable name", idseqsingle, idseqmulti);
   DONE();
 
 // ELSE annotatedseq

src/libponyc/ast/token.h

@@ -73,7 +73,6 @@ typedef enum token_id
   TK_QUESTION,
   TK_UNARY_MINUS,
   TK_ELLIPSIS,
-  TK_DONTCARE,
   TK_CONSTANT,
 
   // Newline symbols, only used by lexer and parser
@@ -103,6 +102,7 @@ typedef enum token_id
   TK_VAR,
   TK_LET,
   TK_EMBED,
+  TK_DONTCARE,
   TK_NEW,
   TK_FUN,
   TK_BE,
@@ -192,6 +192,7 @@ typedef enum token_id
   TK_THISTYPE,
   TK_FUNTYPE,
   TK_LAMBDATYPE,
+  TK_DONTCARETYPE,
   TK_INFERTYPE,
   TK_ERRORTYPE,
 
@@ -221,6 +222,7 @@ typedef enum token_id
   TK_CASES,
   TK_CASE,
   TK_MATCH_CAPTURE,
+  TK_MATCH_DONTCARE,
 
   TK_REFERENCE,
   TK_PACKAGEREF,
@@ -236,6 +238,7 @@ typedef enum token_id
   TK_VARREF,
   TK_LETREF,
   TK_PARAMREF,
+  TK_DONTCAREREF,
   TK_NEWAPP,
   TK_BEAPP,
   TK_FUNAPP,

src/libponyc/ast/treecheckdef.h

@@ -171,7 +171,7 @@ RULE(asop,
 
 RULE(tuple,
   HAS_TYPE(type)
-  ONE_OR_MORE(rawseq, dont_care),
+  ONE_OR_MORE(rawseq),
   TK_TUPLE);
 
 RULE(consume,
@@ -319,7 +319,7 @@ RULE(match_case,
   CHILD(rawseq, none),  // Body
   TK_CASE);
 
-RULE(no_case_expr, HAS_TYPE(dont_care), TK_NONE);
+RULE(no_case_expr, HAS_TYPE(dontcare_type), TK_NONE);
 
 RULE(try_expr,
   HAS_TYPE(type)
@@ -403,8 +403,8 @@ RULE(local_ref,
 
 GROUP(type,
   type_infix, type_tuple, type_arrow, type_this, cap, nominal,
-  type_param_ref, dont_care, fun_type, error_type, lambda_type,
-  literal_type, opliteral_type, control_type, dont_care);
+  type_param_ref, dontcare_type, fun_type, error_type, lambda_type,
+  literal_type, opliteral_type, control_type);
 
 RULE(type_infix, ONE_OR_MORE(type), TK_UNIONTYPE, TK_ISECTTYPE);
 
@@ -458,7 +458,7 @@ RULE(borrowed, LEAF, TK_BORROWED);
 RULE(cap, LEAF, TK_ISO, TK_TRN, TK_REF, TK_VAL, TK_BOX, TK_TAG);
 RULE(control_type, LEAF, TK_IF, TK_CASES, TK_COMPILE_INTRINSIC,
   TK_COMPILE_ERROR, TK_RETURN, TK_BREAK, TK_CONTINUE, TK_ERROR);
-RULE(dont_care, HAS_TYPE(type), TK_DONTCARE);
+RULE(dontcare_type, LEAF, TK_DONTCARETYPE);
 RULE(ellipsis, LEAF, TK_ELLIPSIS);
 RULE(ephemeral, LEAF, TK_EPHEMERAL);
 RULE(error_type, LEAF, TK_ERRORTYPE);