Commit cf749893b592c342cd21fe524c7c45456876cae3 | ix

CHANGELOG.md created+8

...	@@ -0,0 +1,8 @@
1	## 0.1.1-alpha
2
3	- introduce `Report` and show errors during Transform.
4	- add flag --dump-ir for enabling ir dumps
5
6	## 0.1.0-alpha
7
8	initial demo of compiler pipeline in zig.

demo.txt created+1

...	@@ -0,0 +1 @@
1	no "waay

src/compiler/Ast.zig+131-83

...	@@ -34,96 +34,54 @@	...	@@ -34,96 +34,54 @@
34	//! The `Ast` owns token and node data, but not the source code.	34	//! The `Ast` owns token and node data, but not the source code.
35		35
36	/// Externally owned source code.	36	/// Externally owned source code.
37	utf8_source: []const u8,	37	utf8_source: [:0]const u8,
38		38
39	tokens: Token.List.Slice,	39	tokens: Token.List.Slice,
40	nodes: Node.List.Slice,	40	nodes: Node.List.Slice,
41	extra: []const u32,	41	extra: []const u32,
42		42
43	pub fn parse(gpa: std.mem.Allocator, source: []const u8) !Ast {	43	const Parse = @import("Parse.zig");
44	const Parse = @import("Parse.zig");	44	pub fn parse(
45		45	gpa: std.mem.Allocator,
46	var todo: ByteOffset = 0;	46	source: [:0]const u8,
47	var tokens = try Tokenizer.tokenize(source, gpa, &todo);	47	ast_error: *Parse.AstError,
		48	) Parse.Error!Ast {
		49	var err_byte: ByteOffset = 0;
		50	var tokens = Tokenizer.tokenize(
		51	source,
		52	gpa,
		53	&err_byte,
		54	) catch \|err\| switch (err) {
		55	error.OutOfMemory => return error.OutOfMemory,
		56	else => {
		57	ast_error.* = .{ .tokenization = .{
		58	.err = err,
		59	.byte = err_byte,
		60	} };
		61	return error.ParseFail;
		62	},
		63	};
48	errdefer tokens.deinit(gpa);	64	errdefer tokens.deinit(gpa);
49		65
50	var parser = Parse.init(gpa, source, tokens);	66	var parser = Parse.init(gpa, source, tokens);
51	errdefer parser.deinit();	67	errdefer parser.deinit();
52		68
53	try parser.nodes.ensureUnusedCapacity(gpa, @divFloor(tokens.len, 4));	69	try parser.nodes.ensureUnusedCapacity(
		70	gpa,
		71	@max(1, @divFloor(tokens.len, 4)),
		72	);
54		73
55	try parser.parseRoot();	74	parser.parseRoot() catch \|err\| switch (err) {
		75	error.OutOfMemory => return error.OutOfMemory,
		76	error.ParseFail => {
		77	ast_error.* = parser.ast_error.?;
		78	return error.ParseFail;
		79	},
		80	};
56		81
57	return try parser.finalize(tokens);	82	return try parser.finalize(tokens);
58	}	83	}
59		84
60	pub const Location = struct {
61	line: usize,
62	column: usize,
63	line_start: usize,
64	line_end: usize,
65	};
66
67	pub const Span = struct {
68	start: u32,
69	end: u32,
70	main: u32,
71	};
72
73	// This can be removed when the following issue is implemented:
74	// See: https://github.com/ziglang/zig/issues/104
75	// See: https://github.com/ziglang/zig/issues/3806
76	pub const OptionalNodeIndex = enum(u32) {
77	none = 0,
78	_,
79
80	pub fn init(idx: ?NodeIndex) OptionalNodeIndex {
81	assert(idx != 0);
82	return @enumFromInt(idx orelse return .none);
83	}
84
85	pub fn unwrap(idx: OptionalNodeIndex) ?NodeIndex {
86	return switch (idx) {
87	.none => null,
88	_ => @intFromEnum(idx),
89	};
90	}
91	};
92
93	pub const OptionalTokenIndex = enum(u32) {
94	none = 0,
95	_,
96
97	pub fn init(idx: ?TokenIndex) OptionalTokenIndex {
98	assert(idx != 0);
99	return @enumFromInt(idx orelse return .none);
100	}
101
102	pub fn unwrap(idx: OptionalTokenIndex) ?TokenIndex {
103	return switch (idx) {
104	.none => null,
105	_ => @intFromEnum(idx),
106	};
107	}
108	};
109
110	pub const OptionalExtraIndex = enum(u32) {
111	none = std.math.maxInt(u32),
112	_,
113
114	pub fn init(idx: ?ExtraIndex) OptionalExtraIndex {
115	assert(idx != std.math.maxInt(u32));
116	return @enumFromInt(idx orelse return .none);
117	}
118
119	pub fn unwrap(idx: OptionalExtraIndex) ?ExtraIndex {
120	return switch (idx) {
121	.none => null,
122	_ => @intFromEnum(idx),
123	};
124	}
125	};
126
127	pub const Node = struct {	85	pub const Node = struct {
128	tag: Tag,	86	tag: Tag,
129	main_token: TokenIndex,	87	main_token: TokenIndex,
...	@@ -462,6 +420,76 @@ pub const Node = struct {	...	@@ -462,6 +420,76 @@ pub const Node = struct {
462	};	420	};
463	};	421	};
464		422
		423	pub const Error = error{ OutOfMemory, ParseFail };
		424
		425	pub const Location = struct {
		426	// TODO: add an ordinal number type, replace with Ordinal(usize)
		427	line: usize,
		428	column: usize,
		429	line_start: usize,
		430	line_end: usize,
		431	};
		432
		433	pub const Span = struct {
		434	start: u32,
		435	end: u32,
		436	main: u32,
		437	};
		438
		439	// This can be removed when the following issue is implemented:
		440	// See: https://github.com/ziglang/zig/issues/104
		441	// See: https://github.com/ziglang/zig/issues/3806
		442	pub const OptionalNodeIndex = enum(u32) {
		443	none = 0,
		444	_,
		445
		446	pub fn init(idx: ?NodeIndex) OptionalNodeIndex {
		447	assert(idx != 0);
		448	return @enumFromInt(idx orelse return .none);
		449	}
		450
		451	pub fn unwrap(idx: OptionalNodeIndex) ?NodeIndex {
		452	return switch (idx) {
		453	.none => null,
		454	_ => @intFromEnum(idx),
		455	};
		456	}
		457	};
		458
		459	pub const OptionalTokenIndex = enum(u32) {
		460	none = 0,
		461	_,
		462
		463	pub fn init(idx: ?TokenIndex) OptionalTokenIndex {
		464	assert(idx != 0);
		465	return @enumFromInt(idx orelse return .none);
		466	}
		467
		468	pub fn unwrap(idx: OptionalTokenIndex) ?TokenIndex {
		469	return switch (idx) {
		470	.none => null,
		471	_ => @intFromEnum(idx),
		472	};
		473	}
		474	};
		475
		476	pub const OptionalExtraIndex = enum(u32) {
		477	none = std.math.maxInt(u32),
		478	_,
		479
		480	pub fn init(idx: ?ExtraIndex) OptionalExtraIndex {
		481	assert(idx != std.math.maxInt(u32));
		482	return @enumFromInt(idx orelse return .none);
		483	}
		484
		485	pub fn unwrap(idx: OptionalExtraIndex) ?ExtraIndex {
		486	return switch (idx) {
		487	.none => null,
		488	_ => @intFromEnum(idx),
		489	};
		490	}
		491	};
		492
465	pub const SubList = struct {	493	pub const SubList = struct {
466	/// Either NodeIndex or TokenIndex, depending on where this came from.	494	/// Either NodeIndex or TokenIndex, depending on where this came from.
467	start: u32,	495	start: u32,
...	@@ -598,6 +626,19 @@ pub fn fullParamDecl(ast: Ast, node: NodeIndex) full.ParamDecl {	...	@@ -598,6 +626,19 @@ pub fn fullParamDecl(ast: Ast, node: NodeIndex) full.ParamDecl {
598	};	626	};
599	}	627	}
600		628
		629	pub fn idToken(ast: Ast, node: NodeIndex) TokenIndex {
		630	return switch (ast.nodeTag(node)) {
		631	.param_decl,
		632	.id,
		633	.label,
		634	=> ast.nodeToken(node),
		635	.fn_decl,
		636	.var_decl,
		637	=> ast.nodeToken(node) + 1,
		638	else => unreachable, // does not have an id
		639	};
		640	}
		641
601	pub fn nodeStatements(ast: *const Ast, node: NodeIndex) []const NodeIndex {	642	pub fn nodeStatements(ast: *const Ast, node: NodeIndex) []const NodeIndex {
602	const data = ast.nodes.items(.data)[node].list;	643	const data = ast.nodes.items(.data)[node].list;
603	return data.slice(ast);	644	return data.slice(ast);
...	@@ -605,15 +646,15 @@ pub fn nodeStatements(ast: *const Ast, node: NodeIndex) []const NodeIndex {	...	@@ -605,15 +646,15 @@ pub fn nodeStatements(ast: *const Ast, node: NodeIndex) []const NodeIndex {
605		646
606	pub fn tokenLocation(self: Ast, start_offset: ByteOffset, token_index: TokenIndex) Location {	647	pub fn tokenLocation(self: Ast, start_offset: ByteOffset, token_index: TokenIndex) Location {
607	var loc = Location{	648	var loc = Location{
608	.line = 0,	649	.line = 1,
609	.column = 0,	650	.column = 1,
610	.line_start = start_offset,	651	.line_start = start_offset,
611	.line_end = self.source.len,	652	.line_end = self.utf8_source.len,
612	};	653	};
613	const token_start = self.tokens.items(.start)[token_index];	654	const token_start = self.tokens.items(.offset)[token_index];
614		655
615	// Scan to by line until we go past the token start	656	// Scan to by line until we go past the token start
616	while (std.mem.indexOfScalarPos(u8, self.source, loc.line_start, '\n')) \|i\| {	657	while (std.mem.indexOfScalarPos(u8, self.utf8_source, loc.line_start, '\n')) \|i\| {
617	if (i >= token_start) {	658	if (i >= token_start) {
618	break; // Went past	659	break; // Went past
619	}	660	}
...	@@ -622,17 +663,17 @@ pub fn tokenLocation(self: Ast, start_offset: ByteOffset, token_index: TokenInde	...	@@ -622,17 +663,17 @@ pub fn tokenLocation(self: Ast, start_offset: ByteOffset, token_index: TokenInde
622	}	663	}
623		664
624	const offset = loc.line_start;	665	const offset = loc.line_start;
625	for (self.source[offset..], 0..) \|c, i\| {	666	for (self.utf8_source[offset..], 0..) \|c, i\| {
626	if (i + offset == token_start) {	667	if (i + offset == token_start) {
627	loc.line_end = i + offset;	668	loc.line_end = i + offset;
628	while (loc.line_end < self.source.len and self.source[loc.line_end] != '\n') {	669	while (loc.line_end < self.utf8_source.len and self.utf8_source[loc.line_end] != '\n') {
629	loc.line_end += 1;	670	loc.line_end += 1;
630	}	671	}
631	return loc;	672	return loc;
632	}	673	}
633	if (c == '\n') {	674	if (c == '\n') {
634	loc.line += 1;	675	loc.line += 1;
635	loc.column = 0;	676	loc.column = 1;
636	loc.line_start = i + 1;	677	loc.line_start = i + 1;
637	} else {	678	} else {
638	loc.column += 1;	679	loc.column += 1;
...	@@ -651,6 +692,11 @@ pub fn basicNodeSlice(ast: Ast, node: NodeIndex) []const u8 {	...	@@ -651,6 +692,11 @@ pub fn basicNodeSlice(ast: Ast, node: NodeIndex) []const u8 {
651	return ast.tokenSlice(token);	692	return ast.tokenSlice(token);
652	}	693	}
653		694
		695	pub fn fnDeclIdToken(tree: Ast, node: NodeIndex) TokenIndex {
		696	assert(tree.nodeTag(node) == .fn_decl);
		697	return tree.nodeToken(node) + 1;
		698	}
		699
654	pub fn extraData(tree: Ast, index: usize, comptime T: type) T {	700	pub fn extraData(tree: Ast, index: usize, comptime T: type) T {
655	const fields = std.meta.fields(T);	701	const fields = std.meta.fields(T);
656	var result: T = undefined;	702	var result: T = undefined;
...	@@ -699,3 +745,5 @@ const Tokenizer = @import("Tokenizer.zig");	...	@@ -699,3 +745,5 @@ const Tokenizer = @import("Tokenizer.zig");
699	const Token = Tokenizer.Token;	745	const Token = Tokenizer.Token;
700	const TokenList = Tokenizer.Token.List;	746	const TokenList = Tokenizer.Token.List;
701	const ByteOffset = Tokenizer.ByteOffset;	747	const ByteOffset = Tokenizer.ByteOffset;
		748
		749	const Report = @import("Report.zig");

src/compiler/Comp.zig+60-11

 @@ -10,6 +10,8 @@ files: std.StringArrayHashMapUnmanaged(*File) = .{},
 /// # of sema instructions in this compilation cycle
 analyzed_instructions: usize = 0,
+options: Options = .{},
 // TODO: consider computing this dynamicly per-link
 current_compile_decls: std.ArrayListUnmanaged(Decl.Index) = .{},
 @@ -74,7 +76,16 @@ pub const File = struct {
     path_digest: ?Cache.BinDigest,
-    pub fn getSource(file: *File, gpa: Allocator) !Source {
+    const GetSourceError =
+        std.fs.File.OpenError ||
+        std.fs.File.ReadError ||
+        std.fs.File.StatError ||
+        Allocator.Error || error{
+        FileTooBig,
+        UnexpectedEndOfFile,
+    };
+    pub fn getSource(file: *File, gpa: Allocator) GetSourceError!Source {
         if (file.source) |source| {
             return source;
         }
 @@ -111,17 +122,42 @@ pub const File = struct {
         return loaded_source;
     }
-    pub fn getAst(file: *File, gpa: Allocator) !Ast {
+    const GetAstError = GetSourceError || Parse.Error;
+    pub fn getAst(file: *File, comp: *Comp) GetAstError!Ast {
         if (file.ast) |ast| {
             return ast;
         }
-        const ast = try Ast.parse(gpa, (try file.getSource(gpa)).code);
+        var ast_err: Parse.AstError = undefined;
+        const ast = Ast.parse(
+            comp.gpa,
+            (try file.getSource(comp.gpa)).code,
+            &ast_err,
+        ) catch |err| switch (err) {
+            error.OutOfMemory => return err,
+            error.ParseFail => {
+                var report = try ast_err.toReport(comp, file);
+                //  TODO: let this merge with other report
+                report.dumpToStdErr(comp) catch {};
+                report.deinit(comp.gpa);
+                return err;
+            },
+        };
+        if (comp.options.dump_ir) {
+            ast.dumpToWriter(std.io.getStdErr().writer()) catch {
+                // Simplify the error set.
+            };
+        }
         file.ast = ast;
         return ast;
     }
-    pub fn getTeg(file: *File, gpa: Allocator) !Teg {
+    const GetTegError = GetAstError || Transform.Error;
+    pub fn getTeg(file: *File, comp: *Comp, fail_report: *Report) GetTegError!Teg {
         if (file.teg) |teg| {
             return teg;
         }
 @@ -136,7 +172,13 @@ pub const File = struct {
         };
         file.path_digest = bin_digest;
-        const teg = try Transform.transform(try file.getAst(gpa), gpa);
+        const teg = try Transform.transform(
+            try file.getAst(comp),
+            comp.gpa,
+            file,
+            fail_report,
+        );
         file.teg = teg;
         return teg;
     }
 @@ -290,8 +332,6 @@ pub const Namespace = struct {
 pub fn indexTegModule(comp: *Comp, file: *File, module_index: Teg.Inst.Index) !Namespace.Index {
     const teg = file.teg.?;
-    log.info("Index {} @ {d}", .{ std.fmt.fmtSliceHexLower(file.path_digest.?[0..8]), @intFromEnum(module_index) });
     const ns = try comp.intern_pool.createNamespace(comp.gpa, .{
         .parent = .none,
         .file_scope = file,
 @@ -322,7 +362,13 @@ pub fn indexTegModule(comp: *Comp, file: *File, module_index: Teg.Inst.Index) !N
 }
 pub fn semaEntryPoint(comp: *Comp, file: *File) !void {
-    const teg = try file.getTeg(comp.gpa);
+    var report: Report = undefined;
+    const teg = try file.getTeg(comp, &report);
+    // TODO, move this elsewhere. we should print all errors and warnings at once
+    if (teg.report.warnings.len > 0) {
+        try teg.report.dumpToStdErr(comp);
+    }
     const ns = try comp.indexTegModule(file, .main_module_inst);
     const decl_index = comp.intern_pool.namespacePtr(ns).lookupDirect(comp, .empty) orelse
 @@ -354,9 +400,6 @@ pub fn codeGenDecl(c: *Comp, decl: *Decl, air: *const Air) !void {
     const t = tracer.trace(@src());
     defer t.end();
-    std.debug.print("Start Codegen\n", .{});
-    defer std.debug.print("End Codegen\n", .{});
     var arena_state = std.heap.ArenaAllocator.init(c.gpa);
     defer arena_state.deinit();
     const arena = arena_state.allocator();
 @@ -371,6 +414,10 @@ pub fn link(c: *Comp, writer: anytype) !void {
     try JsLink.linkJsCode(c, c.current_compile_decls.items, writer);
 }
+const Options = struct {
+    dump_ir: bool = false,
+};
 const Comp = @This();
 const std = @import("std");
 @@ -391,5 +438,7 @@ const Transform = @import("Transform.zig");
 const JsNodeGen = @import("JsNodeGen.zig");
 const JsTree = @import("JsTree.zig");
 const JsLink = @import("JsLink.zig");
+const Report = @import("Report.zig");
+const Parse = @import("Parse.zig");
 const tracer = @import("tracer");

src/compiler/JsLink.zig+1-4

src/compiler/JsNodeGen.zig-2

...	@@ -117,7 +117,6 @@ pub fn build(c: *JsNodeGen) !void {	...	@@ -117,7 +117,6 @@ pub fn build(c: *JsNodeGen) !void {
117	}	117	}
118		118
119	fn codeGenInst(c: *JsNodeGen, inst: Air.Inst.Index) Error!void {	119	fn codeGenInst(c: *JsNodeGen, inst: Air.Inst.Index) Error!void {
120	std.debug.print("JsNodeGen: #{d}\n", .{@intFromEnum(inst)});
121	switch (c.air.instTag(inst)) {	120	switch (c.air.instTag(inst)) {
122	.arg => unreachable, // codegen must skip over these instructions.	121	.arg => unreachable, // codegen must skip over these instructions.
123		122
...	@@ -400,7 +399,6 @@ fn codeGenBranch(c: *JsNodeGen, inst: Air.Inst.Ref, temporary: ?Node.Index) !?Br	...	@@ -400,7 +399,6 @@ fn codeGenBranch(c: *JsNodeGen, inst: Air.Inst.Ref, temporary: ?Node.Index) !?Br
400	this_label,	399	this_label,
401	});	400	});
402		401
403	std.debug.print("requires_label {}\n", .{requires_label});
404	const block: Air.IndexList = c.air.instData(.block, i);	402	const block: Air.IndexList = c.air.instData(.block, i);
405	for (block.slice(c.air)) \|item\| {	403	for (block.slice(c.air)) \|item\| {
406	try c.codeGenInst(item);	404	try c.codeGenInst(item);

src/compiler/Parse.zig+62-27

...	@@ -1,7 +1,7 @@	...	@@ -1,7 +1,7 @@
1	//! Parser takes a Token list and creates a Node list. It does not own the source code or	1	//! Parser takes a Token list and creates a Node list. It does not own the source code or
2	//! token data, however transferring into an `Ast` does move token list ownership.	2	//! token data, however transferring into an `Ast` does move token list ownership.
3	gpa: Allocator,	3	gpa: Allocator,
4	source: []const u8,	4	source: [:0]const u8,
5	token_tags: []const Token.Tag,	5	token_tags: []const Token.Tag,
6	token_starts: []const ByteOffset,	6	token_starts: []const ByteOffset,
7		7
...	@@ -12,7 +12,7 @@ scratch: ArrayListUnmanaged(NodeIndex) = .{},	...	@@ -12,7 +12,7 @@ scratch: ArrayListUnmanaged(NodeIndex) = .{},
12		12
13	ast_error: ?AstError = null,	13	ast_error: ?AstError = null,
14		14
15	pub fn init(gpa: Allocator, source: []const u8, tokens: Token.List.Slice) Parse {	15	pub fn init(gpa: Allocator, source: [:0]const u8, tokens: Token.List.Slice) Parse {
16	return Parse{	16	return Parse{
17	.gpa = gpa,	17	.gpa = gpa,
18	.source = source,	18	.source = source,
...	@@ -36,8 +36,6 @@ pub fn finalize(p: *Parse, token_slice: Token.List.Slice) !Ast {	...	@@ -36,8 +36,6 @@ pub fn finalize(p: *Parse, token_slice: Token.List.Slice) !Ast {
36	.extra = extra,	36	.extra = extra,
37	};	37	};
38		38
39	ast.dumpToWriter(std.io.getStdErr().writer()) catch {};
40
41	return ast;	39	return ast;
42	}	40	}
43		41
...	@@ -50,18 +48,52 @@ pub fn deinit(p: *Parse) void {	...	@@ -50,18 +48,52 @@ pub fn deinit(p: *Parse) void {
50	}	48	}
51		49
52	pub const Error = error{	50	pub const Error = error{
53	SyntaxError,	51	OutOfMemory,
		52	ParseFail,
54	};	53	};
55		54
56	/// When error.SyntaxError is returned, `p.err` is set to this as well as `p.pos`	55	/// When error.SyntaxError is returned, `p.err` is set to this as well as `p.pos`
57	pub const AstError = union(enum) {	56	pub const AstError = union(enum) {
		57	tokenization: struct {
		58	err: Tokenizer.TokenizeError,
		59	byte: u32,
		60	},
58	expected_token: Token.Tag,	61	expected_token: Token.Tag,
59	expected_statement,	62	expected_statement,
60	expected_expr,	63	expected_expr,
61	expected_else_value,	64	expected_else_value,
		65
		66	pub fn toReport(parse_err: AstError, comp: Comp, file: Comp.File) !Report {
		67	return (Report.Err{ .ast = parse_err }).toReport(comp, file);
		68	}
		69
		70	pub fn build(err: AstError, ctx: *Report.MsgContext) !void {
		71	switch (err) {
		72	.tokenization => \|data\| {
		73	// TODO: consider making tokenizer emit AstError more directly,
		74	// so that it can just annotate start and ends
		75	try ctx.msg(switch (data.err) {
		76	error.UnterminatedString => .{ .codepoint_eol = data.byte },
		77	else => .{ .codepoint = data.byte },
		78	}, "{s}", .{switch (data.err) {
		79	error.OutOfMemory => unreachable,
		80	error.InvalidSyntax => "unexpected token",
		81	error.TrailingNumericSeparator => "trailing numberic separator",
		82	error.UnterminatedString => "unterminated string literal",
		83	error.InvalidEscapeSequence => "invalid escape sequence",
		84	error.InvalidEscapeCodepoint => "invalid escape codepoint",
		85	error.InvalidUtf8 => "file is not valid utf8",
		86	error.GitConflict => "git merge conflict encountered",
		87	}});
		88	},
		89
		90	else => unreachable,
		91	//
		92	}
		93	}
62	};	94	};
63		95
64	pub fn parseRoot(p: *Parse) !void {	96	pub fn parseRoot(p: *Parse) Error!void {
65	const trace = t.trace(@src());	97	const trace = t.trace(@src());
66	defer trace.end();	98	defer trace.end();
67		99
...	@@ -82,7 +114,7 @@ pub fn parseRoot(p: *Parse) !void {	...	@@ -82,7 +114,7 @@ pub fn parseRoot(p: *Parse) !void {
82	};	114	};
83	}	115	}
84		116
85	fn parseModuleMembers(p: *Parse) !SubList {	117	fn parseModuleMembers(p: *Parse) Error!SubList {
86	const trace = t.trace(@src());	118	const trace = t.trace(@src());
87	defer trace.end();	119	defer trace.end();
88		120
...	@@ -103,7 +135,7 @@ fn parseModuleMembers(p: *Parse) !SubList {	...	@@ -103,7 +135,7 @@ fn parseModuleMembers(p: *Parse) !SubList {
103	.kw_enum,	135	.kw_enum,
104	.kw_trait,	136	.kw_trait,
105	.kw_impl,	137	.kw_impl,
106	=> return error.Todo,	138	=> \|tag\| std.debug.panic("TODO: parse {s}", .{@tagName(tag)}),
107		139
108	.eof => break,	140	.eof => break,
109		141
...	@@ -120,7 +152,7 @@ fn parseModuleMembers(p: *Parse) !SubList {	...	@@ -120,7 +152,7 @@ fn parseModuleMembers(p: *Parse) !SubList {
120	return p.scratchToSubList(scratch_top);	152	return p.scratchToSubList(scratch_top);
121	}	153	}
122		154
123	fn parseModuleDecl(p: *Parse) !NodeIndex {	155	fn parseModuleDecl(p: *Parse) Error!NodeIndex {
124	const trace = t.trace(@src());	156	const trace = t.trace(@src());
125	defer trace.end();	157	defer trace.end();
126		158
...	@@ -135,7 +167,7 @@ fn parseModuleDecl(p: *Parse) !NodeIndex {	...	@@ -135,7 +167,7 @@ fn parseModuleDecl(p: *Parse) !NodeIndex {
135	};	167	};
136	}	168	}
137		169
138	fn parseFnDecl(p: *Parse) !NodeIndex {	170	fn parseFnDecl(p: *Parse) Error!NodeIndex {
139	const trace = t.trace(@src());	171	const trace = t.trace(@src());
140	defer trace.end();	172	defer trace.end();
141		173
...	@@ -170,7 +202,7 @@ fn parseFnDecl(p: *Parse) !NodeIndex {	...	@@ -170,7 +202,7 @@ fn parseFnDecl(p: *Parse) !NodeIndex {
170	});	202	});
171	}	203	}
172		204
173	fn parseParamList(p: *Parse) !SubList {	205	fn parseParamList(p: *Parse) Error!SubList {
174	const trace = t.trace(@src());	206	const trace = t.trace(@src());
175	defer trace.end();	207	defer trace.end();
176		208
...	@@ -199,7 +231,7 @@ fn parseParamList(p: *Parse) !SubList {	...	@@ -199,7 +231,7 @@ fn parseParamList(p: *Parse) !SubList {
199	return p.scratchToSubList(scratch_top);	231	return p.scratchToSubList(scratch_top);
200	}	232	}
201		233
202	fn parseParamDecl(p: *Parse) !NodeIndex {	234	fn parseParamDecl(p: *Parse) Error!NodeIndex {
203	const trace = t.trace(@src());	235	const trace = t.trace(@src());
204	defer trace.end();	236	defer trace.end();
205		237
...	@@ -218,7 +250,7 @@ fn parseParamDecl(p: *Parse) !NodeIndex {	...	@@ -218,7 +250,7 @@ fn parseParamDecl(p: *Parse) !NodeIndex {
218	});	250	});
219	}	251	}
220		252
221	fn parseBlock(p: *Parse) !NodeIndex {	253	fn parseBlock(p: *Parse) Error!NodeIndex {
222	const trace = t.trace(@src());	254	const trace = t.trace(@src());
223	defer trace.end();	255	defer trace.end();
224		256
...	@@ -236,7 +268,7 @@ fn parseBlock(p: *Parse) !NodeIndex {	...	@@ -236,7 +268,7 @@ fn parseBlock(p: *Parse) !NodeIndex {
236	});	268	});
237	}	269	}
238		270
239	fn parseLabelExpression(p: *Parse) !NodeIndex {	271	fn parseLabelExpression(p: *Parse) Error!NodeIndex {
240	const trace = t.trace(@src());	272	const trace = t.trace(@src());
241	defer trace.end();	273	defer trace.end();
242		274
...	@@ -252,7 +284,7 @@ fn parseLabelExpression(p: *Parse) !NodeIndex {	...	@@ -252,7 +284,7 @@ fn parseLabelExpression(p: *Parse) !NodeIndex {
252	});	284	});
253	}	285	}
254		286
255	fn parseBlockContents(p: *Parse) !struct { Node.Tag, SubList } {	287	fn parseBlockContents(p: *Parse) Error!struct { Node.Tag, SubList } {
256	const trace = t.trace(@src());	288	const trace = t.trace(@src());
257	defer trace.end();	289	defer trace.end();
258		290
...	@@ -308,7 +340,7 @@ fn parseBlockContents(p: *Parse) !struct { Node.Tag, SubList } {	...	@@ -308,7 +340,7 @@ fn parseBlockContents(p: *Parse) !struct { Node.Tag, SubList } {
308	}	340	}
309		341
310	/// VarDecl <- (kw_pub)? (kw_var \| kw_let) id (colon expr)? equal expr semi	342	/// VarDecl <- (kw_pub)? (kw_var \| kw_let) id (colon expr)? equal expr semi
311	fn parseVarDecl(p: *Parse) !NodeIndex {	343	fn parseVarDecl(p: *Parse) Error!NodeIndex {
312	const trace = t.trace(@src());	344	const trace = t.trace(@src());
313	defer trace.end();	345	defer trace.end();
314		346
...	@@ -521,7 +553,7 @@ const Operator = enum(u8) {	...	@@ -521,7 +553,7 @@ const Operator = enum(u8) {
521	}	553	}
522	};	554	};
523		555
524	fn parseExpression(p: *Parse) anyerror!NodeIndex {	556	fn parseExpression(p: *Parse) Error!NodeIndex {
525	const trace = t.trace(@src());	557	const trace = t.trace(@src());
526	defer trace.end();	558	defer trace.end();
527		559
...	@@ -543,7 +575,7 @@ fn parseExpression(p: *Parse) anyerror!NodeIndex {	...	@@ -543,7 +575,7 @@ fn parseExpression(p: *Parse) anyerror!NodeIndex {
543	return expr;	575	return expr;
544	}	576	}
545		577
546	fn parseExpressionPart(p: *Parse) !NodeIndex {	578	fn parseExpressionPart(p: *Parse) Error!NodeIndex {
547	const trace = t.trace(@src());	579	const trace = t.trace(@src());
548	defer trace.end();	580	defer trace.end();
549		581
...	@@ -644,7 +676,7 @@ fn parseExpressionPart(p: *Parse) !NodeIndex {	...	@@ -644,7 +676,7 @@ fn parseExpressionPart(p: *Parse) !NodeIndex {
644	return expr;	676	return expr;
645	}	677	}
646		678
647	fn parseBinarySuffix(p: *Parse, lhs: NodeIndex, op: Operator) !NodeIndex {	679	fn parseBinarySuffix(p: *Parse, lhs: NodeIndex, op: Operator) Error!NodeIndex {
648	const trace = t.trace(@src());	680	const trace = t.trace(@src());
649	defer trace.end();	681	defer trace.end();
650		682
...	@@ -673,7 +705,7 @@ fn parseBinarySuffix(p: *Parse, lhs: NodeIndex, op: Operator) !NodeIndex {	...	@@ -673,7 +705,7 @@ fn parseBinarySuffix(p: *Parse, lhs: NodeIndex, op: Operator) !NodeIndex {
673	});	705	});
674	}	706	}
675		707
676	fn parseReturn(p: *Parse) !NodeIndex {	708	fn parseReturn(p: *Parse) Error!NodeIndex {
677	const trace = t.trace(@src());	709	const trace = t.trace(@src());
678	defer trace.end();	710	defer trace.end();
679		711
...	@@ -693,7 +725,7 @@ fn parseReturn(p: *Parse) !NodeIndex {	...	@@ -693,7 +725,7 @@ fn parseReturn(p: *Parse) !NodeIndex {
693	});	725	});
694	}	726	}
695		727
696	fn parseBreak(p: *Parse) !NodeIndex {	728	fn parseBreak(p: *Parse) Error!NodeIndex {
697	const trace = t.trace(@src());	729	const trace = t.trace(@src());
698	defer trace.end();	730	defer trace.end();
699		731
...	@@ -719,7 +751,7 @@ fn parseBreak(p: *Parse) !NodeIndex {	...	@@ -719,7 +751,7 @@ fn parseBreak(p: *Parse) !NodeIndex {
719	});	751	});
720	}	752	}
721		753
722	fn parseBuiltinCall(p: *Parse) !NodeIndex {	754	fn parseBuiltinCall(p: *Parse) Error!NodeIndex {
723	const trace = t.trace(@src());	755	const trace = t.trace(@src());
724	defer trace.end();	756	defer trace.end();
725		757
...	@@ -736,7 +768,7 @@ fn parseBuiltinCall(p: *Parse) !NodeIndex {	...	@@ -736,7 +768,7 @@ fn parseBuiltinCall(p: *Parse) !NodeIndex {
736	});	768	});
737	}	769	}
738		770
739	fn parseCallArgs(p: *Parse) anyerror!SubList {	771	fn parseCallArgs(p: *Parse) Error!SubList {
740	const trace = t.trace(@src());	772	const trace = t.trace(@src());
741	defer trace.end();	773	defer trace.end();
742		774
...	@@ -764,7 +796,7 @@ fn parseCallArgs(p: *Parse) anyerror!SubList {	...	@@ -764,7 +796,7 @@ fn parseCallArgs(p: *Parse) anyerror!SubList {
764	return p.scratchToSubList(scratch_top);	796	return p.scratchToSubList(scratch_top);
765	}	797	}
766		798
767	fn parseIf(p: *Parse) !NodeIndex {	799	fn parseIf(p: *Parse) Error!NodeIndex {
768	const trace = t.trace(@src());	800	const trace = t.trace(@src());
769	defer trace.end();	801	defer trace.end();
770		802
...	@@ -795,7 +827,7 @@ fn parseIf(p: *Parse) !NodeIndex {	...	@@ -795,7 +827,7 @@ fn parseIf(p: *Parse) !NodeIndex {
795	});	827	});
796	}	828	}
797		829
798	fn parseElseValue(p: *Parse) anyerror!NodeIndex {	830	fn parseElseValue(p: *Parse) Error!NodeIndex {
799	const trace = t.trace(@src());	831	const trace = t.trace(@src());
800	defer trace.end();	832	defer trace.end();
801		833
...	@@ -810,7 +842,7 @@ fn parseElseValue(p: *Parse) anyerror!NodeIndex {	...	@@ -810,7 +842,7 @@ fn parseElseValue(p: *Parse) anyerror!NodeIndex {
810	};	842	};
811	}	843	}
812		844
813	fn parseGroup(p: *Parse) !NodeIndex {	845	fn parseGroup(p: *Parse) Error!NodeIndex {
814	const trace = t.trace(@src());	846	const trace = t.trace(@src());
815	defer trace.end();	847	defer trace.end();
816		848
...	@@ -855,7 +887,7 @@ fn addExtra(p: *Parse, comptime T: type, extra: T) !ExtraIndex {	...	@@ -855,7 +887,7 @@ fn addExtra(p: *Parse, comptime T: type, extra: T) !ExtraIndex {
855	fn fail(p: *Parse, ast_error: AstError) Error {	887	fn fail(p: *Parse, ast_error: AstError) Error {
856	p.ast_error = ast_error;	888	p.ast_error = ast_error;
857	std.debug.print("got {s}", .{@tagName(p.currentTag())});	889	std.debug.print("got {s}", .{@tagName(p.currentTag())});
858	return Error.SyntaxError;	890	return Error.ParseFail;
859	}	891	}
860		892
861	fn blankNode(p: *Parse, comptime tag: Token.Tag) !NodeIndex {	893	fn blankNode(p: *Parse, comptime tag: Token.Tag) !NodeIndex {
...	@@ -959,3 +991,6 @@ const optionalNodeIndex = OptionalNodeIndex.init;	...	@@ -959,3 +991,6 @@ const optionalNodeIndex = OptionalNodeIndex.init;
959	const optionalExtraIndex = OptionalExtraIndex.init;	991	const optionalExtraIndex = OptionalExtraIndex.init;
960		992
961	const t = @import("tracer");	993	const t = @import("tracer");
		994
		995	const Report = @import("Report.zig");
		996	const Comp = @import("Comp.zig");

src/compiler/Report.zig+466-15

...	@@ -1,38 +1,489 @@	...	@@ -1,38 +1,489 @@
		1	//! A Report is a list of errors and warnings for a part of a compilation.
		2	//!
		3	//! `Err/Warn` structs are unions of every possible message, the logic
		4	//! for printing them is all done here.
		5	//!
		6	//! Across the compiler, `Report.Builder`s' `.err` and `.warn` methods
		7	//! can add messages by ID.
1	errors: []Item = &.{},	8	errors: []Item = &.{},
2	warns: []Item = &.{},	9	warnings: []Item = &.{},
3	string_bytes: []u8 = &.{},	10	string_bytes: []u8 = &.{},
4		11
5	pub const Item = struct {	12	pub const Item = struct {
6	offset: ByteOffset,	13	/// What is this diagnostic
7	message: String,	14	data: union { @"error": Err, warn: Warn },
		15	/// What file was it in
		16	url: String,
		17
		18	/// Type is never stored, inferred from which list it is in.
		19	pub const Type = enum { @"error", warn };
		20	};
		21
		22	pub const Err = union(enum) {
		23	pub const FieldEnum = std.meta.FieldEnum(Err);
		24
		25	/// Constructs a `Report` with just a single error.
		26	pub fn toReport(err: Err, comp: Comp, file: File) !Report {
		27	const string_bytes = try comp.gpa.dupe(u8, file.file_path);
		28	errdefer comp.gpa.free(string_bytes);
		29
		30	const errors = try comp.gpa.dupe(Item, &.{.{
		31	.data = .{ .@"error" = err },
		32	.url = .{ .start = 0, .len = @intCast(string_bytes.len) },
		33	}});
		34	errdefer comp.gpa.free(errors);
		35
		36	return .{ .errors = errors, .string_bytes = string_bytes };
		37	}
		38
		39	ast: @import("Parse.zig").AstError,
		40
		41	unknown_identifier: struct {
		42	node: NodeIndex,
		43
		44	fn build(data: @This(), ctx: *MsgContext) !void {
		45	try ctx.msgNode(data.node, "unknown identifier '{s}'", .{
		46	ctx.ast.basicNodeSlice(data.node),
		47	});
		48	}
		49	},
		50	invalid_builtin: struct {
		51	node: NodeIndex,
		52
		53	fn build(data: @This(), ctx: *MsgContext) !void {
		54	try ctx.msgToken(
		55	ctx.ast.nodeToken(data.node),
		56	"unknown built-in function '{s}'",
		57	.{ctx.ast.basicNodeSlice(data.node)},
		58	);
		59	}
		60	},
		61	invalid_discard: struct {
		62	node: NodeIndex,
		63
		64	fn build(data: @This(), ctx: *MsgContext) !void {
		65	try ctx.msgNode(data.node, "local cannot be named '_'", .{});
		66	}
		67	},
		68	duplicate_function: struct {
		69	node: NodeIndex,
		70	prev: NodeIndex,
		71
		72	fn build(data: @This(), ctx: *MsgContext) !void {
		73	try ctx.msgToken(ctx.ast.fnDeclIdToken(data.node), "re-declaration of '{s}'", .{
		74	ctx.ast.fullFnDecl(data.node).id,
		75	});
		76	try ctx.msgToken(ctx.ast.fnDeclIdToken(data.prev), "also declared here", .{});
		77	// TODO: be able to store a full list of other declarations
		78	// and add them here as "and here" and so on
		79	}
		80	},
		81	duplicate_param: struct {
		82	node: NodeIndex,
		83	prev: NodeIndex,
		84
		85	fn build(data: @This(), ctx: *MsgContext) !void {
		86	try ctx.msgToken(ctx.ast.fnDeclIdToken(data.node), "re-declaration of '{s}'", .{
		87	ctx.ast.fullParamDecl(data.node).id,
		88	});
		89	try ctx.msgToken(ctx.ast.nodeToken(data.prev), "also declared here", .{});
		90	// TODO: be able to store a full list of other declarations
		91	// and add them here as "and here" and so on
		92	}
		93	},
		94	invalid_return: struct {
		95	node: NodeIndex,
		96
		97	fn build(data: @This(), ctx: *MsgContext) !void {
		98	try ctx.msgNode(
		99	data.node,
		100	"return cannot be used here",
		101	.{},
		102	);
		103	}
		104	},
		105	unresolved_label: struct {
		106	node: NodeIndex,
		107
		108	fn build(data: @This(), ctx: *MsgContext) !void {
		109	try ctx.msg(
		110	.{ .two_tokens = ctx.ast.nodeToken(data.node) + 1 },
		111	"unknown block label '{s}'",
		112	.{ctx.ast.tokenSlice(ctx.ast.nodeToken(data.node) + 2)},
		113	);
		114	}
		115	},
		116	shadow_builtin_variable: struct {
		117	node: NodeIndex,
		118
		119	fn build(data: @This(), ctx: *MsgContext) !void {
		120	try ctx.msgToken(
		121	ctx.ast.nodeToken(data.node) + 1,
		122	"declaration shadows built-in constant '{s}'",
		123	.{ctx.ast.fullVarDecl(data.node).id},
		124	);
		125	}
		126	},
		127	};
		128
		129	pub const Warn = union(enum) {
		130	pub const FieldEnum = std.meta.FieldEnum(Warn);
		131
		132	unused: struct {
		133	node: NodeIndex,
		134	kind: UnusedKind,
		135
		136	fn build(data: @This(), ctx: *MsgContext) !void {
		137	try ctx.msgToken(ctx.ast.idToken(data.node), "unused {s} '{s}'", .{
		138	@tagName(data.kind),
		139	switch (ctx.ast.nodeTag(data.node)) {
		140	.param_decl => ctx.ast.fullParamDecl(data.node).id,
		141	.fn_decl => ctx.ast.fullFnDecl(data.node).id,
		142	.var_decl => ctx.ast.fullVarDecl(data.node).id,
		143	else => unreachable, // not a valid symbol node
		144	},
		145	});
		146	}
		147	},
		148	};
		149
		150	pub const UnusedKind = enum {
		151	param,
		152	constant,
		153	variable,
		154	function,
8	};	155	};
9		156
10	pub const String = struct {	157	pub const String = struct {
11	start: u32,	158	start: u32,
12	len: u32,	159	len: u32,
13		160
14	pub fn message(item: Item, report: Report) []const u8 {	161	pub fn slice(str: String, report: Report) []const u8 {
15	return report.string_bytes[item.message_start..][0..item.message_len];	162	return report.string_bytes[str.start..][0..str.len];
		163	}
		164	};
		165
		166	const Prefix = enum {
		167	@"error",
		168	warn,
		169	note,
		170
		171	fn color(prefix: Prefix) std.io.tty.Color {
		172	return switch (prefix) {
		173	.@"error" => .red,
		174	.warn => .yellow,
		175	.note => .blue,
		176	};
16	}	177	}
17	};	178	};
18		179
19	pub fn hasErrors(report: Report) bool {	180	pub fn deinit(report: Report, gpa: Allocator) void {
20	return report.errors.len > 0;	181	gpa.free(report.errors);
		182	gpa.free(report.warnings);
		183	gpa.free(report.string_bytes);
21	}	184	}
22		185
23	pub fn hasWarns(report: Report) bool {	186	pub fn dumpToWriter(report: Report, comp: *Comp, writer: anytype) !void {
24	return report.warns.len > 0;	187	const tty = util.detectTTYFromWriter(writer);
		188	try writer.print("{d} errors, {d} warnings\n", .{ report.errors.len, report.warnings.len });
		189	for (report.errors) \|item\| {
		190	try report.formatItem(item, .@"error", comp, writer, tty);
		191	}
		192	for (report.warnings) \|item\| {
		193	try report.formatItem(item, .warn, comp, writer, tty);
		194	}
25	}	195	}
26		196
27	pub const Wip = struct {	197	pub fn formatItem(
28	errors: ArrayListUnmanaged(Item),	198	report: Report,
29	warns: ArrayListUnmanaged(Item),	199	item: Item,
30	string_bytes: ArrayListUnmanaged(u8),	200	item_type: Item.Type,
		201	comp: *Comp,
		202	writer: anytype,
		203	tty: std.io.tty.Config,
		204	) !void {
		205	var arena = std.heap.ArenaAllocator.init(comp.gpa);
		206	defer arena.deinit();
		207	var sfa = std.heap.stackFallback(16384, arena.allocator());
		208	const alloc = sfa.get();
		209
		210	const file = try comp.getFile(item.url.slice(report));
		211
		212	const need_load_ast = !(item_type == .@"error" and item.data.@"error" == .ast);
		213	const ast_data = if (need_load_ast) try file.getAst(comp) else undefined;
		214	const ast = if (need_load_ast) &ast_data else null;
		215
		216	var ctx: MsgContext = .{
		217	.arena = alloc,
		218	.comp = comp,
		219	.file = file,
		220	.ast = &ast_data,
		221	.components = ArrayListUnmanaged(Component).initCapacity(alloc, 16) catch
		222	unreachable, // stack-fallback allocator has enough space.
		223	};
		224
		225	switch (item_type) {
		226	inline else => \|tag\| switch (@field(item.data, @tagName(tag))) {
		227	inline else => \|payload\| try payload.build(&ctx),
		228	},
		229	}
		230
		231	assert(ctx.components.items.len > 0);
		232
		233	for (ctx.components.items, 1..) \|component, i\| {
		234	const prefix: Prefix = if (i == 1)
		235	switch (item_type) {
		236	inline else => \|tag\| @field(Prefix, @tagName(tag)),
		237	}
		238	else
		239	.note;
		240
		241	const is_last = i == ctx.components.items.len;
		242
		243	if (i != 1) {
		244	try tty.setColor(writer, .dim);
		245	try writer.writeAll(" │ ");
		246	try tty.setColor(writer, .reset);
		247	}
		248
		249	try tty.setColor(writer, prefix.color());
		250	try writer.writeAll(@tagName(prefix));
		251	try tty.setColor(writer, .reset);
		252	try tty.setColor(writer, .dim);
		253	try writer.writeAll(": ");
		254	try tty.setColor(writer, .reset);
		255	try writer.writeAll(component.message_utf8);
		256	try writer.writeAll("\n");
		257
		258	const SliceLoc = struct {
		259	loc: Location,
		260	byte: ByteOffset,
		261	slice: []const u8,
		262	};
		263
		264	const utf8_source = (try file.getSource(comp.gpa)).code;
		265
		266	const at: SliceLoc = switch (component.highlight) {
		267	inline .node,
		268	.token,
		269	.two_tokens,
		270	=> \|payload, tag\| loc: {
		271	const token_index: TokenIndex = switch (tag) {
		272	.node => ast.?.nodeToken(payload),
		273	.token, .two_tokens => payload,
		274	else => comptime unreachable,
		275	};
		276
		277	const location = ast.?.tokenLocation(0, token_index);
		278	const token = ast.?.tokens.get(token_index);
		279
		280	break :loc .{
		281	.loc = location,
		282	.slice = token.slice(ast.?.utf8_source),
		283	.byte = token.offset,
		284	};
		285	},
		286
		287	inline .codepoint,
		288	.codepoint_eol,
		289	=> \|b, tag\| loc: {
		290	const len = switch (tag) {
		291	.codepoint => std.unicode.utf8ByteSequenceLength(utf8_source[b]) catch 1,
		292	.codepoint_eol => if (std.mem.indexOfScalar(u8, utf8_source[b..], '\n')) \|n\|
		293	n
		294	else
		295	(utf8_source.len - b),
		296	else => comptime unreachable,
		297	};
		298	break :loc .{
		299	.byte = b,
		300	.slice = utf8_source[b..][0..len],
		301	.loc = .{
		302	.column = 0,
		303	.line = 0,
		304	.line_start = 0,
		305	.line_end = utf8_source.len,
		306	},
		307	};
		308	},
		309	};
		310
		311	const line_before = utf8_source[at.loc.line_start..at.byte];
		312	const trimmed_before = std.mem.trimLeft(u8, line_before, " \t\r");
		313	const line_after = utf8_source[at.byte + at.slice.len .. at.loc.line_end + 1];
		314	const width = std.fmt.count("{d}", .{at.loc.line});
		315
		316	try tty.setColor(writer, .dim);
		317	try writer.writeByteNTimes(' ', width);
		318	try writer.print("{s}{s}:{d}:{d}\n", .{
		319	if (i == 1) "┬ " else "├ ",
		320	file.file_path,
		321	at.loc.line,
		322	at.loc.column,
		323	});
		324	try writer.print("{d}", .{at.loc.line});
		325	try writer.print("{s} ", .{"│"});
		326	try tty.setColor(writer, .reset);
		327	try writer.writeAll(trimmed_before);
		328	try tty.setColor(writer, prefix.color());
		329	try tty.setColor(writer, .bold);
		330	try writer.writeAll(at.slice);
		331	try tty.setColor(writer, .reset);
		332	try writer.writeAll(line_after);
		333	try tty.setColor(writer, .dim);
		334	try writer.writeAll(if (is_last) " ╰" else " ├");
		335	try writer.writeBytesNTimes(
		336	"─",
		337	width +
		338	(at.byte - at.loc.line_start) -
		339	(line_before.len - trimmed_before.len),
		340	);
		341	try tty.setColor(writer, .reset);
		342	try tty.setColor(writer, prefix.color());
		343	try tty.setColor(writer, .bold);
		344	try writer.writeBytesNTimes("━", at.slice.len);
		345	try tty.setColor(writer, .reset);
		346	try writer.writeAll("\n");
		347	}
		348	}
		349
		350	pub fn dumpToStdErr(report: Report, comp: *Comp) !void {
		351	const stderr = std.io.getStdErr().writer();
		352
		353	// Disabled because of coloring issues.
		354	// var buf = std.io.bufferedWriter(stderr);
		355	// const writer = buf.writer();
		356	try dumpToWriter(report, comp, stderr);
		357	// try buf.flush();
		358	}
		359
		360	pub const Builder = struct {
		361	gpa: Allocator,
		362	current_file: *Comp.File,
		363	errors: ArrayListUnmanaged(Item) = .{},
		364	warnings: ArrayListUnmanaged(Item) = .{},
		365	string_bytes: ArrayListUnmanaged(u8) = .{},
		366	current_file_str: ?String = null,
		367
		368	pub fn err(b: *Builder, comptime kind: Err.FieldEnum, data: std.meta.FieldType(Err, kind)) !void {
		369	try b.errors.append(b.gpa, .{
		370	.data = .{ .@"error" = @unionInit(Err, @tagName(kind), data) },
		371	.url = try b.currentFileString(),
		372	});
		373	}
		374
		375	pub fn warn(b: *Builder, comptime kind: Warn.FieldEnum, data: std.meta.FieldType(Warn, kind)) !void {
		376	try b.warnings.append(b.gpa, .{
		377	.data = .{ .warn = @unionInit(Warn, @tagName(kind), data) },
		378	.url = try b.currentFileString(),
		379	});
		380	}
		381
		382	pub fn addString(b: *Builder, str: []const u8) !String {
		383	const top = b.string_bytes.items.len;
		384	try b.string_bytes.appendSlice(b.gpa, str);
		385	return .{ .start = @intCast(top), .len = @intCast(str.len) };
		386	}
		387
		388	pub fn currentFileString(b: *Builder) !String {
		389	return b.current_file_str orelse {
		390	b.current_file_str = try b.addString(b.current_file.file_path);
		391	return b.current_file_str.?;
		392	};
		393	}
		394
		395	pub fn deinit(b: *Builder) void {
		396	b.errors.deinit(b.gpa);
		397	b.warnings.deinit(b.gpa);
		398	}
		399
		400	pub fn finalize(b: *Builder) !Report {
		401	defer b.* = undefined;
		402	errdefer b.deinit();
		403
		404	const errors = try b.errors.toOwnedSlice(b.gpa);
		405	errdefer b.gpa.free(errors);
		406
		407	const warnings = try b.warnings.toOwnedSlice(b.gpa);
		408	errdefer b.gpa.free(warnings);
		409
		410	const string_bytes = try b.string_bytes.toOwnedSlice(b.gpa);
		411	errdefer b.gpa.free(string_bytes);
		412
		413	return .{
		414	.errors = errors,
		415	.warnings = warnings,
		416	.string_bytes = string_bytes,
		417	};
		418	}
		419	};
		420
		421	pub const Highlight = union(enum) {
		422	// Highlight one token
		423	token: TokenIndex,
		424	// Highlight two tokens
		425	two_tokens: TokenIndex,
		426	// Highlight a Node
		427	node: NodeIndex,
		428	// Highlight a codepoint
		429	codepoint: ByteOffset,
		430	// Highlight from the start of a codepoint to '\n'
		431	codepoint_eol: ByteOffset,
		432	};
		433
		434	const Component = struct {
		435	message_utf8: []const u8,
		436	highlight: Highlight,
		437	};
		438
		439	/// This serves as a small API to make creating messages easier.
		440	pub const MsgContext = struct {
		441	file: *File,
		442	comp: *Comp,
		443	ast: *const Ast,
		444
		445	/// This arena is is a stack fallback with 16kb, then
		446	/// falling back to an arena allocator. This should
		447	/// make the cost of arbitrarily creating `Component`
		448	/// nodes extremely lightweight.
		449	arena: Allocator,
		450
		451	/// First component is "error" or "warn"
		452	/// Trailing components are "note"
		453	components: ArrayListUnmanaged(Component),
		454
		455	pub fn msg(m: *MsgContext, highlight: Highlight, comptime fmt: []const u8, args: anytype) !void {
		456	const text = try std.fmt.allocPrint(m.arena, fmt, args);
		457	try m.components.append(m.arena, .{
		458	.message_utf8 = text,
		459	.highlight = highlight,
		460	});
		461	}
		462
		463	pub fn msgNode(m: *MsgContext, node: NodeIndex, comptime fmt: []const u8, args: anytype) !void {
		464	return m.msg(.{ .node = node }, fmt, args);
		465	}
		466
		467	pub fn msgToken(m: *MsgContext, token: TokenIndex, comptime fmt: []const u8, args: anytype) !void {
		468	return m.msg(.{ .token = token }, fmt, args);
		469	}
31	};	470	};
32		471
33	const Report = @This();	472	const Report = @This();
34	const std = @import("std");	473	const std = @import("std");
35	const ArrayListUnmanaged = std.ArrayListUnmanaged;	474	const ArrayListUnmanaged = std.ArrayListUnmanaged;
		475	const Allocator = std.mem.Allocator;
36	const assert = std.debug.assert;	476	const assert = std.debug.assert;
37	pub const Tokenizer = @import("Tokenizer.zig");	477
38	pub const ByteOffset = Tokenizer.ByteOffset;	478	const Tokenizer = @import("Tokenizer.zig");
		479	const ByteOffset = Tokenizer.ByteOffset;
		480
		481	const Ast = @import("Ast.zig");
		482	const TokenIndex = Ast.TokenIndex;
		483	const NodeIndex = Ast.NodeIndex;
		484	const Location = Ast.Location;
		485
		486	const Comp = @import("Comp.zig");
		487	const File = Comp.File;
		488
		489	const util = @import("util.zig");

src/compiler/Sema.zig+11-32

 @@ -37,7 +37,6 @@ const InstMap = struct {
     pub fn init(allocator: Allocator, code: *const Teg, inst_start: Teg.Inst.Index) !InstMap {
         const decl = code.declData(inst_start);
-        log.info("pre-allocating {d} map-indicies for InstMap", .{decl.instruction_count});
         const insts = try allocator.alloc(Air.Inst.Ref.Optional, decl.instruction_count);
         @memset(insts, .none);
         return .{
 @@ -72,7 +71,6 @@ pub fn init(s: *Sema) !void {
     try s.comp.current_compile_decls.append(s.gpa, s.decl_index);
-    log.info("init sema on index {}", .{s.decl_index});
     s.inst_map = try InstMap.init(
         s.arena,
         &s.code,
 @@ -193,8 +191,6 @@ fn analyzeCode(s: *Sema, in: Teg.Inst.Index) !Air.Inst.Ref {
     const t = tracer.traceFmt(@src(), "%{d}, {s}", .{ @intFromEnum(in), @tagName(s.code.instTag(in)) });
     defer t.end();
-    log.info("runtime %{d} = .{s}", .{ @intFromEnum(in), @tagName(s.code.instTag(in)) });
     const out = try s.analyzeCodeInner(in);
     if (s.code.instTag(in) != .block)
         try s.inst_map.add(in, out);
 @@ -226,8 +222,6 @@ fn analyzeCodeInner(s: *Sema, in: Teg.Inst.Index) !Air.Inst.Ref {
             const string = rhs.asString(s.ip);
             const base_type = s.typeOfRef(base);
-            log.info("process field with type {any} . {s}", .{ base_type, string.toSlice(s.ip) });
             switch (base_type) {
                 .opaque_type => {
                     return s.addInstRef(.field, .{
 @@ -277,18 +271,12 @@ fn analyzeCodeInner(s: *Sema, in: Teg.Inst.Index) !Air.Inst.Ref {
             }
             const left_type = s.typeOfRef(lhs);
+            _ = left_type; // autofix
             const right_type = s.typeOfRef(rhs);
+            _ = right_type; // autofix
             // TODO: validate type for operation
-            log.info("runtime {s} operation on {} and {}", .{
-                @tagName(tag),
-                left_type.toKey(s.ip),
-                right_type.toKey(s.ip),
-            });
-            // todo: type check
             return s.addInstRef(tag.airTag(), .{
                 .lhs = lhs,
                 .rhs = rhs,
 @@ -300,14 +288,10 @@ fn analyzeCodeInner(s: *Sema, in: Teg.Inst.Index) !Air.Inst.Ref {
             const operand = try s.resolveRefRuntime(un.operand);
             const operand_type = s.typeOfRef(operand);
+            _ = operand_type; // autofix
             // TODO: validate type for operation
-            log.info("runtime {s} operation on {}", .{
-                @tagName(tag),
-                operand_type.toKey(s.ip),
-            });
             return s.addInstRef(tag.airTag(), .{
                 .operand = operand,
             });
 @@ -425,7 +409,6 @@ fn analyzeCodeInner(s: *Sema, in: Teg.Inst.Index) !Air.Inst.Ref {
 fn markInst(s: *Sema, inst: Air.Inst.Index) void {
     const flags: *Air.Flags = &s.air_instructions.items(.flags)[@intFromEnum(inst)];
     flags.uses = flags.uses.increment();
-    log.info("markInst {}", .{inst});
 }
 fn analyzeRefToBlock(s: *Sema, in: Teg.Inst.Ref) anyerror!Air.Inst.Ref {
 @@ -485,15 +468,13 @@ fn resolveRefComptime(s: *Sema, ref: Teg.Inst.Ref) !Value {
         }
     }
-    const tag = s.code.instTag(in);
+    switch (s.code.instTag(in)) {
-    log.info("comptime %{d} = .{s}", .{ @intFromEnum(in), @tagName(tag) });
-    switch (tag) {
         .string => {
             const string: Teg.SlicedString = s.code.instPayload(.string, in);
             return s.makeValue(.{ .string = string.slice(s.code.string_bytes) });
         },
-        else => std.debug.panic("TODO: comptime analyze of .{s}", .{@tagName(tag)}),
+        else => |tag| std.debug.panic("TODO: comptime analyze of .{s}", .{@tagName(tag)}),
     }
 }
 @@ -509,12 +490,13 @@ fn resolveDecl(s: *Sema, decl_index: Decl.Index) !Value {
             var arena = std.heap.ArenaAllocator.init(s.gpa);
             defer arena.deinit();
+            var todo: Report = undefined;
             var sema: Sema = .{
                 .comp = s.comp,
                 .gpa = s.gpa,
                 .ip = s.ip,
                 .arena = arena.allocator(),
-                .code = try s.ip.namespacePtr(decl.src_namespace).file_scope.getTeg(s.gpa),
+                .code = try s.ip.namespacePtr(decl.src_namespace).file_scope.getTeg(s.comp, &todo),
                 .decl = decl,
                 .decl_index = decl_index,
                 .namespace_index = decl.src_namespace,
 @@ -529,10 +511,9 @@ fn resolveDecl(s: *Sema, decl_index: Decl.Index) !Value {
             return value;
         },
-        .complete => {
+        else => {
             return decl.val;
         },
-        else => |tag| std.debug.panic("TODO: load_decl when decl is in state .{s}", .{@tagName(tag)}),
     }
 }
 @@ -573,7 +554,6 @@ const IncompleteBlock = struct {
 };
 fn addBlock(s: *Sema) !IncompleteBlock {
-    log.info("start block at @{d}", .{s.air_instructions.len});
     try s.air_instructions.append(s.gpa, .{
         .tag = .block,
         .data = undefined,
 @@ -583,7 +563,6 @@ fn addBlock(s: *Sema) !IncompleteBlock {
 }
 fn addInstRef(s: *Sema, comptime tag: Air.Inst.Tag, data: tag.Data()) !Air.Inst.Ref {
-    log.info("add air .{s}, {}", .{ @tagName(tag), data });
     try s.air_instructions.append(s.gpa, .{
         .tag = tag,
         .data = Air.EncodedData.encode(tag, data),
 @@ -627,11 +606,9 @@ fn mapTeg(s: *Sema, inst: Teg.Inst.Ref) Air.Inst.Ref {
 }
 fn validateType(s: *Sema, checked_type: Value, wanted_type: InternPool.Index) !void {
+    _ = s; // autofix
     if (checked_type.toIntern() != wanted_type) {
-        log.info("failed type check {}, got {}", .{ Value.fromInterned(wanted_type).fmt(s.ip), checked_type.fmt(s.ip) });
         return error.TypeCheckFailed;
-    } else {
-        log.info("passed type check {}", .{checked_type.fmt(s.ip)});
     }
 }
 @@ -690,4 +667,6 @@ const InternPool = @import("InternPool.zig");
 const NullTerminatedString = InternPool.NullTerminatedString;
 const dod = @import("dod.zig");
+const Report = @import("Report.zig");
 const tracer = @import("tracer");

src/compiler/Teg.zig+4

...	@@ -3,6 +3,7 @@	...	@@ -3,6 +3,7 @@
3	insts: Inst.List.Slice,	3	insts: Inst.List.Slice,
4	string_bytes: []const u8,	4	string_bytes: []const u8,
5	extra: ExtraStore = .{},	5	extra: ExtraStore = .{},
		6	report: Report,
6		7
7	pub const ExtraStore = dod.ExtraStore(.teg, .@"const");	8	pub const ExtraStore = dod.ExtraStore(.teg, .@"const");
8	pub const EncodedData = dod.EncodedData(Inst.Tag, u64);	9	pub const EncodedData = dod.EncodedData(Inst.Tag, u64);
...	@@ -343,6 +344,7 @@ pub fn deinit(teg: *Teg, gpa: std.mem.Allocator) void {	...	@@ -343,6 +344,7 @@ pub fn deinit(teg: *Teg, gpa: std.mem.Allocator) void {
343	gpa.free(teg.string_bytes);	344	gpa.free(teg.string_bytes);
344	teg.extra.deinit(gpa);	345	teg.extra.deinit(gpa);
345	teg.insts.deinit(gpa);	346	teg.insts.deinit(gpa);
		347	teg.report.deinit(gpa);
346	teg.* = undefined;	348	teg.* = undefined;
347	}	349	}
348		350
...	@@ -433,3 +435,5 @@ const Float = dod.Float;	...	@@ -433,3 +435,5 @@ const Float = dod.Float;
433	const InternPool = @import("InternPool.zig");	435	const InternPool = @import("InternPool.zig");
434		436
435	const Air = @import("Air.zig");	437	const Air = @import("Air.zig");
		438
		439	const Report = @import("Report.zig");

src/compiler/Tokenizer.zig+96-79

...	@@ -5,18 +5,14 @@	...	@@ -5,18 +5,14 @@
5	//! Retriving payloads like string/numeric values, identifier names, etc is	5	//! Retriving payloads like string/numeric values, identifier names, etc is
6	//! done separately, either in Parse or Transform, when the respective data	6	//! done separately, either in Parse or Transform, when the respective data
7	//! is actually needed. Tokenization does guarante that such data is valid.	7	//! is actually needed. Tokenization does guarante that such data is valid.
8		8	code: [:0]const u8,
9	// TODO: make `code` a [:0]const u8. this doesnt disallow null bytes, but
10	// it requires the source to end with one. This can make byteAt(1) return u8
11	// instead of an optional. I am told this improves performance.
12	code: []const u8,
13	pos: ByteOffset = 0,	9	pos: ByteOffset = 0,
14		10
15	/// Tokenize source code into a list of tokens.	11	/// Tokenize source code into a list of tokens.
16	/// When an error occurs, the position of the error is returned in `error_pos`.	12	/// When an error occurs, the position of the error is returned in `error_pos`.
17	/// Caller owns returned slice.	13	/// Caller owns returned slice.
18	pub fn tokenize(	14	pub fn tokenize(
19	source_code: []const u8,	15	source_code: [:0]const u8,
20	gpa: mem.Allocator,	16	gpa: mem.Allocator,
21	error_pos: *ByteOffset,	17	error_pos: *ByteOffset,
22	) TokenizeError!Token.List.Slice {	18	) TokenizeError!Token.List.Slice {
...	@@ -233,7 +229,7 @@ pub const Token = struct {	...	@@ -233,7 +229,7 @@ pub const Token = struct {
233	///	229	///
234	/// If you are computing the length of something that is knowable	230	/// If you are computing the length of something that is knowable
235	/// without the source code, use token.tag.staticLen() instead.	231	/// without the source code, use token.tag.staticLen() instead.
236	pub fn len(token: Token, utf8_source: []const u8) u32 {	232	pub fn len(token: Token, utf8_source: [:0]const u8) u32 {
237	return token.tag.staticLen() orelse {	233	return token.tag.staticLen() orelse {
238	var tokenizer = Tokenizer{ .code = utf8_source, .pos = token.offset };	234	var tokenizer = Tokenizer{ .code = utf8_source, .pos = token.offset };
239	assert(	235	assert(
...	@@ -246,7 +242,7 @@ pub const Token = struct {	...	@@ -246,7 +242,7 @@ pub const Token = struct {
246	};	242	};
247	}	243	}
248		244
249	pub fn slice(token: Token, utf8_source: []const u8) []const u8 {	245	pub fn slice(token: Token, utf8_source: [:0]const u8) []const u8 {
250	return utf8_source[token.offset..][0..token.len(utf8_source)];	246	return utf8_source[token.offset..][0..token.len(utf8_source)];
251	}	247	}
252		248
...	@@ -294,7 +290,7 @@ pub fn next(t: *Tokenizer) !?Token {	...	@@ -294,7 +290,7 @@ pub fn next(t: *Tokenizer) !?Token {
294	const trace = T.trace(@src());	290	const trace = T.trace(@src());
295	defer trace.end();	291	defer trace.end();
296		292
297	switch (t.byte() orelse return null) {	293	switch (t.byte()) {
298	' ', '\t', '\n', '\r' => unreachable, // call skipWhitespace	294	' ', '\t', '\n', '\r' => unreachable, // call skipWhitespace
299		295
300	'a'...'z', 'A'...'Z', '_', '$' => {	296	'a'...'z', 'A'...'Z', '_', '$' => {
...	@@ -303,7 +299,7 @@ pub fn next(t: *Tokenizer) !?Token {	...	@@ -303,7 +299,7 @@ pub fn next(t: *Tokenizer) !?Token {
303	return try t.tokenizeIdentifier(start);	299	return try t.tokenizeIdentifier(start);
304	},	300	},
305		301
306	'@' => switch (t.byteAt(1) orelse 0) {	302	'@' => switch (t.nextByte()) {
307	'"' => {	303	'"' => {
308	const start = t.pos;	304	const start = t.pos;
309	t.pos += 1;	305	t.pos += 1;
...	@@ -319,12 +315,12 @@ pub fn next(t: *Tokenizer) !?Token {	...	@@ -319,12 +315,12 @@ pub fn next(t: *Tokenizer) !?Token {
319		315
320	'0'...'9' => return try t.tokenizeNumber(),	316	'0'...'9' => return try t.tokenizeNumber(),
321		317
322	'.' => switch (t.byteAt(1) orelse 0) {	318	'.' => switch (t.nextByte()) {
323	'0'...'9' => return try t.tokenizeNumber(),	319	'0'...'9' => return try t.tokenizeNumber(),
324	else => return t.simpleToken(.dot),	320	else => return t.simpleToken(.dot),
325	},	321	},
326		322
327	'/' => switch (t.byteAt(1) orelse 0) {	323	'/' => switch (t.nextByte()) {
328	'=' => return t.simpleToken(.slash_equal),	324	'=' => return t.simpleToken(.slash_equal),
329	'/' => {	325	'/' => {
330	if ((t.byteAt(2) orelse 0 == '/') and	326	if ((t.byteAt(2) orelse 0 == '/') and
...	@@ -349,7 +345,7 @@ pub fn next(t: *Tokenizer) !?Token {	...	@@ -349,7 +345,7 @@ pub fn next(t: *Tokenizer) !?Token {
349	else => return t.simpleToken(.slash),	345	else => return t.simpleToken(.slash),
350	},	346	},
351		347
352	'\\' => switch (t.byteAt(1) orelse 0) {	348	'\\' => switch (t.nextByte()) {
353	'\\' => {	349	'\\' => {
354	const start = t.pos;	350	const start = t.pos;
355	t.pos = @intCast(mem.indexOfScalarPos(u8, t.code, t.pos + 2, '\n') orelse t.code.len);	351	t.pos = @intCast(mem.indexOfScalarPos(u8, t.code, t.pos + 2, '\n') orelse t.code.len);
...	@@ -360,30 +356,30 @@ pub fn next(t: *Tokenizer) !?Token {	...	@@ -360,30 +356,30 @@ pub fn next(t: *Tokenizer) !?Token {
360		356
361	'"' => return try t.tokenizeString(),	357	'"' => return try t.tokenizeString(),
362		358
363	'&' => switch (t.byteAt(1) orelse 0) {	359	'&' => switch (t.nextByte()) {
364	'&' => return t.simpleToken(.and_and),	360	'&' => return t.simpleToken(.and_and),
365	'=' => return t.simpleToken(.and_equal),	361	'=' => return t.simpleToken(.and_equal),
366	else => return t.simpleToken(.@"and"),	362	else => return t.simpleToken(.@"and"),
367	},	363	},
368	'\|' => switch (t.byteAt(1) orelse 0) {	364	'\|' => switch (t.nextByte()) {
369	'\|' => return t.simpleToken(.pipe_pipe),	365	'\|' => return t.simpleToken(.pipe_pipe),
370	'=' => return t.simpleToken(.pipe_equal),	366	'=' => return t.simpleToken(.pipe_equal),
371	else => return t.simpleToken(.pipe),	367	else => return t.simpleToken(.pipe),
372	},	368	},
373	'^' => switch (t.byteAt(1) orelse 0) {	369	'^' => switch (t.nextByte()) {
374	'^' => return t.simpleToken(.caret),	370	'^' => return t.simpleToken(.caret),
375	'=' => return t.simpleToken(.caret_equal),	371	'=' => return t.simpleToken(.caret_equal),
376	else => return t.simpleToken(.caret),	372	else => return t.simpleToken(.caret),
377	},	373	},
378	'!' => switch (t.byteAt(1) orelse 0) {	374	'!' => switch (t.nextByte()) {
379	'=' => return t.simpleToken(.equal_equal),	375	'=' => return t.simpleToken(.equal_equal),
380	else => return t.simpleToken(.bang_equal),	376	else => return t.simpleToken(.bang_equal),
381	},	377	},
382	'=' => switch (t.byteAt(1) orelse 0) {	378	'=' => switch (t.nextByte()) {
383	'=' => return t.simpleToken(.equal_equal),	379	'=' => return t.simpleToken(.equal_equal),
384	else => return t.simpleToken(.equal),	380	else => return t.simpleToken(.equal),
385	},	381	},
386	'>' => switch (t.byteAt(1) orelse 0) {	382	'>' => switch (t.nextByte()) {
387	'=' => return t.simpleToken(.gt_equal),	383	'=' => return t.simpleToken(.gt_equal),
388	'>' => switch (t.byteAt(2) orelse 0) {	384	'>' => switch (t.byteAt(2) orelse 0) {
389	'>' => return t.simpleToken(.gt_gt_gt),	385	'>' => return t.simpleToken(.gt_gt_gt),
...	@@ -391,7 +387,7 @@ pub fn next(t: *Tokenizer) !?Token {	...	@@ -391,7 +387,7 @@ pub fn next(t: *Tokenizer) !?Token {
391	},	387	},
392	else => return t.simpleToken(.gt),	388	else => return t.simpleToken(.gt),
393	},	389	},
394	'<' => switch (t.byteAt(1) orelse 0) {	390	'<' => switch (t.nextByte()) {
395	'=' => return t.simpleToken(.lt_equal),	391	'=' => return t.simpleToken(.lt_equal),
396	'<' => switch (t.byteAt(2) orelse 0) {	392	'<' => switch (t.byteAt(2) orelse 0) {
397	'=' => return t.simpleToken(.lt_lt_equal),	393	'=' => return t.simpleToken(.lt_lt_equal),
...	@@ -399,14 +395,12 @@ pub fn next(t: *Tokenizer) !?Token {	...	@@ -399,14 +395,12 @@ pub fn next(t: *Tokenizer) !?Token {
399	// Git merge conflicts are <<<<<<<	395	// Git merge conflicts are <<<<<<<
400	const start = t.pos;	396	const start = t.pos;
401	t.pos += 2;	397	t.pos += 2;
402	brk: {	398	var remain: u3 = 4;
403	var remain: u3 = 4;	399	while (t.byte() == '<') {
404	while (t.byte() orelse break :brk == '<') {	400	t.pos += 1;
405	t.pos += 1;	401	remain -= 1;
406	remain -= 1;	402	if (remain == 0) {
407	if (remain == 0) {	403	return error.GitConflict;
408	return error.GitConflict;
409	}
410	}	404	}
411	}	405	}
412	t.pos = start;	406	t.pos = start;
...	@@ -415,25 +409,25 @@ pub fn next(t: *Tokenizer) !?Token {	...	@@ -415,25 +409,25 @@ pub fn next(t: *Tokenizer) !?Token {
415	},	409	},
416	else => return t.simpleToken(.lt),	410	else => return t.simpleToken(.lt),
417	},	411	},
418	'+' => switch (t.byteAt(1) orelse 0) {	412	'+' => switch (t.nextByte()) {
419	'=' => return t.simpleToken(.plus_equal),	413	'=' => return t.simpleToken(.plus_equal),
420	else => return t.simpleToken(.plus),	414	else => return t.simpleToken(.plus),
421	},	415	},
422	'-' => switch (t.byteAt(1) orelse 0) {	416	'-' => switch (t.nextByte()) {
423	'=' => return t.simpleToken(.minus_equal),	417	'=' => return t.simpleToken(.minus_equal),
424	'>' => return t.simpleToken(.minus_lt),	418	'>' => return t.simpleToken(.minus_lt),
425	else => return t.simpleToken(.minus),	419	else => return t.simpleToken(.minus),
426	},	420	},
427	'*' => switch (t.byteAt(1) orelse 0) {	421	'*' => switch (t.nextByte()) {
428	'*' => return t.simpleToken(.star_star),	422	'*' => return t.simpleToken(.star_star),
429	'=' => return t.simpleToken(.star_equal),	423	'=' => return t.simpleToken(.star_equal),
430	else => return t.simpleToken(.star),	424	else => return t.simpleToken(.star),
431	},	425	},
432	'%' => switch (t.byteAt(1) orelse 0) {	426	'%' => switch (t.nextByte()) {
433	'=' => return t.simpleToken(.percent_equal),	427	'=' => return t.simpleToken(.percent_equal),
434	else => return t.simpleToken(.percent),	428	else => return t.simpleToken(.percent),
435	},	429	},
436	'~' => switch (t.byteAt(1) orelse 0) {	430	'~' => switch (t.nextByte()) {
437	'=' => return t.simpleToken(.tilde_equal),	431	'=' => return t.simpleToken(.tilde_equal),
438	else => return t.simpleToken(.tilde),	432	else => return t.simpleToken(.tilde),
439	},	433	},
...	@@ -462,19 +456,40 @@ pub fn next(t: *Tokenizer) !?Token {	...	@@ -462,19 +456,40 @@ pub fn next(t: *Tokenizer) !?Token {
462	},	456	},
463		457
464	else => return error.InvalidSyntax,	458	else => return error.InvalidSyntax,
		459
		460	0 => {
		461	if (t.pos == t.code.len) return null;
		462	return error.InvalidSyntax;
		463	},
465	}	464	}
466	}	465	}
467	}	466	}
468		467
469	fn byte(t: Tokenizer) ?u8 {	468	/// Source code is null terminated, so accessing this is safe
470	if (t.pos >= t.code.len) return null;	469	fn byte(t: Tokenizer) u8 {
471	return t.code[t.pos];	470	return t.code[t.pos];
472	}	471	}
473		472
		473	/// Source code is null terminated.
		474	fn nextByte(t: Tokenizer) u8 {
		475	assert(t.byte() < 0x80);
		476	return t.code[t.pos + 1];
		477	}
		478
474	/// Remember that using this only makes sense if you know for certain that all	479	/// Remember that using this only makes sense if you know for certain that all
475	/// previous bytes are ASCII, otherwise you may hit UB	480	/// previous bytes are ASCII, otherwise you may hit UB
476	fn byteAt(t: Tokenizer, offset: u32) ?u8 {	481	fn byteAt(t: Tokenizer, comptime offset: u32) ?u8 {
477	if (t.pos + offset >= t.code.len) return null;	482	if (offset == 1) @compileError("use nextByte()");
		483
		484	// assert that the previous bytes are ASCII
		485	// current byte is allowed to be a continuation byte
		486	for (t.code[t.pos .. t.pos + offset]) \|b\| {
		487	assert(b < 0x80);
		488	}
		489	return t.code[t.pos + offset];
		490	}
		491
		492	fn byteAtRt(t: Tokenizer, offset: u32) ?u8 {
478	// assert that the previous bytes are ASCII	493	// assert that the previous bytes are ASCII
479	// current byte is allowed to be a continuation byte	494	// current byte is allowed to be a continuation byte
480	for (t.code[t.pos .. t.pos + offset]) \|b\| {	495	for (t.code[t.pos .. t.pos + offset]) \|b\| {
...	@@ -484,7 +499,7 @@ fn byteAt(t: Tokenizer, offset: u32) ?u8 {	...	@@ -484,7 +499,7 @@ fn byteAt(t: Tokenizer, offset: u32) ?u8 {
484	}	499	}
485		500
486	fn codePoint(t: Tokenizer) !struct { u3, u21 } {	501	fn codePoint(t: Tokenizer) !struct { u3, u21 } {
487	const b = t.byte() orelse unreachable; // must check the first byte manually	502	const b = t.byte(); // must check the first byte manually
488	assert(b >= 0x80); // must be a continuation byte	503	assert(b >= 0x80); // must be a continuation byte
489		504
490	const byte_len = unicode.utf8ByteSequenceLength(b) catch	505	const byte_len = unicode.utf8ByteSequenceLength(b) catch
...	@@ -498,7 +513,7 @@ fn codePoint(t: Tokenizer) !struct { u3, u21 } {	...	@@ -498,7 +513,7 @@ fn codePoint(t: Tokenizer) !struct { u3, u21 } {
498	}	513	}
499		514
500	fn isAtIdentifierStart(t: *Tokenizer) !bool {	515	fn isAtIdentifierStart(t: *Tokenizer) !bool {
501	switch (t.byte() orelse return false) {	516	switch (t.byte()) {
502	'a'...'z', 'A'...'Z', '_', '$' => return true,	517	'a'...'z', 'A'...'Z', '_', '$' => return true,
503	else => return false,	518	else => return false,
504	0x80...0xFF => {	519	0x80...0xFF => {
...	@@ -513,7 +528,7 @@ fn tokenizeIdentifier(t: *Tokenizer, start_byte: u32) !Token {	...	@@ -513,7 +528,7 @@ fn tokenizeIdentifier(t: *Tokenizer, start_byte: u32) !Token {
513	defer trace.end();	528	defer trace.end();
514		529
515	while (true) {	530	while (true) {
516	switch (t.byte() orelse break) {	531	switch (t.byte()) {
517	'a'...'z', 'A'...'Z', '0'...'9', '_', '$' => {	532	'a'...'z', 'A'...'Z', '0'...'9', '_', '$' => {
518	t.pos += 1;	533	t.pos += 1;
519	},	534	},
...	@@ -555,8 +570,8 @@ fn tokenizeNumber(t: *Tokenizer) !?Token {	...	@@ -555,8 +570,8 @@ fn tokenizeNumber(t: *Tokenizer) !?Token {
555	defer trace.end();	570	defer trace.end();
556		571
557	// Caller ensures the following invariant:	572	// Caller ensures the following invariant:
558	switch (t.byte() orelse unreachable) {	573	switch (t.byte()) {
559	'.' => switch (t.byteAt(1) orelse unreachable) {	574	'.' => switch (t.nextByte()) {
560	'0'...'9' => {},	575	'0'...'9' => {},
561	else => unreachable,	576	else => unreachable,
562	},	577	},
...	@@ -566,7 +581,7 @@ fn tokenizeNumber(t: *Tokenizer) !?Token {	...	@@ -566,7 +581,7 @@ fn tokenizeNumber(t: *Tokenizer) !?Token {
566		581
567	// Parse Hex, Binary, Octal numbers	582	// Parse Hex, Binary, Octal numbers
568	if (t.byte() == '0') {	583	if (t.byte() == '0') {
569	switch (t.byteAt(1) orelse 0) {	584	switch (t.nextByte()) {
570	'x', 'X' => {	585	'x', 'X' => {
571	t.pos += 2;	586	t.pos += 2;
572	try t.consumeIntegerRadix(.hex);	587	try t.consumeIntegerRadix(.hex);
...	@@ -594,8 +609,8 @@ fn tokenizeNumber(t: *Tokenizer) !?Token {	...	@@ -594,8 +609,8 @@ fn tokenizeNumber(t: *Tokenizer) !?Token {
594	try t.consumeInteger();	609	try t.consumeInteger();
595		610
596	// decimal point notation	611	// decimal point notation
597	switch (t.byte() orelse 0) {	612	switch (t.byte()) {
598	'.' => switch (t.byteAt(1) orelse 0) {	613	'.' => switch (t.nextByte()) {
599	'0'...'9' => {	614	'0'...'9' => {
600	t.pos += 1;	615	t.pos += 1;
601	tag = .number_float;	616	tag = .number_float;
...	@@ -611,15 +626,15 @@ fn tokenizeNumber(t: *Tokenizer) !?Token {	...	@@ -611,15 +626,15 @@ fn tokenizeNumber(t: *Tokenizer) !?Token {
611	}	626	}
612	}	627	}
613		628
614	switch (t.byte() orelse 0) {	629	switch (t.byte()) {
615	'e', 'E' => {	630	'e', 'E' => {
616	tag = .number_float;	631	tag = .number_float;
617	t.pos += 1;	632	t.pos += 1;
618		633
619	switch (t.byte() orelse 0) {	634	switch (t.byte()) {
620	'+', '-' => {	635	'+', '-' => {
621	t.pos += 1;	636	t.pos += 1;
622	switch (t.byte() orelse 0) {	637	switch (t.byte()) {
623	'0'...'9' => {},	638	'0'...'9' => {},
624	else => return t.maybeUTF8Error(error.InvalidSyntax),	639	else => return t.maybeUTF8Error(error.InvalidSyntax),
625	}	640	}
...	@@ -631,8 +646,8 @@ fn tokenizeNumber(t: *Tokenizer) !?Token {	...	@@ -631,8 +646,8 @@ fn tokenizeNumber(t: *Tokenizer) !?Token {
631	try t.consumeInteger();	646	try t.consumeInteger();
632		647
633	// decimal point notation	648	// decimal point notation
634	switch (t.byte() orelse 0) {	649	switch (t.byte()) {
635	'.' => switch (t.byteAt(1) orelse 0) {	650	'.' => switch (t.nextByte()) {
636	'0'...'9' => {	651	'0'...'9' => {
637	t.pos += 1;	652	t.pos += 1;
638	tag = .number_float;	653	tag = .number_float;
...	@@ -662,7 +677,7 @@ fn maybeConsumeInteger(t: *Tokenizer) !?void {	...	@@ -662,7 +677,7 @@ fn maybeConsumeInteger(t: *Tokenizer) !?void {
662	const trace = T.trace(@src());	677	const trace = T.trace(@src());
663	defer trace.end();	678	defer trace.end();
664		679
665	switch (t.byte() orelse return null) {	680	switch (t.byte()) {
666	'0'...'9' => {	681	'0'...'9' => {
667	t.pos += 1;	682	t.pos += 1;
668	},	683	},
...	@@ -676,7 +691,7 @@ fn consumeInteger(t: *Tokenizer) !void {	...	@@ -676,7 +691,7 @@ fn consumeInteger(t: *Tokenizer) !void {
676	const trace = T.trace(@src());	691	const trace = T.trace(@src());
677	defer trace.end();	692	defer trace.end();
678		693
679	switch (t.byte() orelse unreachable) {	694	switch (t.byte()) {
680	'0'...'9' => {	695	'0'...'9' => {
681	t.pos += 1;	696	t.pos += 1;
682	},	697	},
...	@@ -684,13 +699,13 @@ fn consumeInteger(t: *Tokenizer) !void {	...	@@ -684,13 +699,13 @@ fn consumeInteger(t: *Tokenizer) !void {
684	}	699	}
685		700
686	while (true) {	701	while (true) {
687	switch (t.byte() orelse return) {	702	switch (t.byte()) {
688	'0'...'9' => {	703	'0'...'9' => {
689	t.pos += 1;	704	t.pos += 1;
690	},	705	},
691	'_' => {	706	'_' => {
692	t.pos += 1;	707	t.pos += 1;
693	switch (t.byte() orelse 0) {	708	switch (t.byte()) {
694	'0'...'9' => {	709	'0'...'9' => {
695	t.pos += 1;	710	t.pos += 1;
696	},	711	},
...	@@ -715,7 +730,7 @@ fn consumeIntegerRadix(t: *Tokenizer, comptime radix: Radix) !void {	...	@@ -715,7 +730,7 @@ fn consumeIntegerRadix(t: *Tokenizer, comptime radix: Radix) !void {
715	};	730	};
716		731
717	first: {	732	first: {
718	const b = t.byte() orelse return;	733	const b = t.byte();
719	inline for (digits) \|digit\| {	734	inline for (digits) \|digit\| {
720	if (b == digit) {	735	if (b == digit) {
721	break :first;	736	break :first;
...	@@ -728,7 +743,7 @@ fn consumeIntegerRadix(t: *Tokenizer, comptime radix: Radix) !void {	...	@@ -728,7 +743,7 @@ fn consumeIntegerRadix(t: *Tokenizer, comptime radix: Radix) !void {
728		743
729	while (true) {	744	while (true) {
730	brk: {	745	brk: {
731	const b = t.byte() orelse return;	746	const b = t.byte();
732	inline for (digits) \|digit\| {	747	inline for (digits) \|digit\| {
733	if (b == digit) {	748	if (b == digit) {
734	break :brk;	749	break :brk;
...	@@ -741,7 +756,7 @@ fn consumeIntegerRadix(t: *Tokenizer, comptime radix: Radix) !void {	...	@@ -741,7 +756,7 @@ fn consumeIntegerRadix(t: *Tokenizer, comptime radix: Radix) !void {
741	}	756	}
742		757
743	fn tokenizeString(t: *Tokenizer) !Token {	758	fn tokenizeString(t: *Tokenizer) !Token {
744	assert(t.byte().? == '"');	759	assert(t.byte() == '"');
745	const start = t.pos;	760	const start = t.pos;
746	try t.consumeString();	761	try t.consumeString();
747	return .{ .tag = .string, .offset = start };	762	return .{ .tag = .string, .offset = start };
...	@@ -751,32 +766,30 @@ fn consumeString(t: *Tokenizer) !void {	...	@@ -751,32 +766,30 @@ fn consumeString(t: *Tokenizer) !void {
751	const trace = T.trace(@src());	766	const trace = T.trace(@src());
752	defer trace.end();	767	defer trace.end();
753		768
754	assert(t.byte().? == '"');	769	const start = t.pos;
		770	assert(t.byte() == '"');
755		771
756	t.pos += 1;	772	t.pos += 1;
757		773
758	while (true) {	774	while (true) {
759	const b = t.byte() orelse return error.UnterminatedString;	775	switch (t.byte()) {
760	switch (b) {
761	'"' => {	776	'"' => {
762	t.pos += 1;	777	t.pos += 1;
763	break;	778	break;
764	},	779	},
765	'\\' => {	780	'\\' => {
766	switch (t.byteAt(1) orelse {	781	switch (t.nextByte()) {
767	// we want error.InvalidEscapeSequence to be at the '\' char,	782	'\n' => {
768	// but we want error.UnterminatedString to be at the end of the source code	783	t.pos = start;
769	t.pos += 1;	784	return error.UnterminatedString;
770	return error.UnterminatedString;	785	},
771	}) {
772	'\n' => return error.UnterminatedString,
773	'\\' => t.pos += 2,	786	'\\' => t.pos += 2,
774	'"' => t.pos += 2,	787	'"' => t.pos += 2,
775	'n' => t.pos += 2,	788	'n' => t.pos += 2,
776	'r' => t.pos += 2,	789	'r' => t.pos += 2,
777	't' => t.pos += 2,	790	't' => t.pos += 2,
778	'0' => switch (t.byteAt(2) orelse {	791	'0' => switch (t.byteAt(2) orelse {
779	t.pos += 2;	792	t.pos = start;
780	return error.UnterminatedString;	793	return error.UnterminatedString;
781	}) {	794	}) {
782	// '\0xFF' hex escape	795	// '\0xFF' hex escape
...	@@ -788,7 +801,7 @@ fn consumeString(t: *Tokenizer) !void {	...	@@ -788,7 +801,7 @@ fn consumeString(t: *Tokenizer) !void {
788	const esc_start = t.pos;	801	const esc_start = t.pos;
789	t.consumeStringUnicodeEscape() catch \|err\| switch (err) {	802	t.consumeStringUnicodeEscape() catch \|err\| switch (err) {
790	error.InvalidEscapeSequence => {	803	error.InvalidEscapeSequence => {
791	if (t.byte().? == '\n')	804	if (t.byte() == '\n')
792	return error.UnterminatedString;	805	return error.UnterminatedString;
793	t.pos = esc_start;	806	t.pos = esc_start;
794	return err;	807	return err;
...	@@ -797,12 +810,16 @@ fn consumeString(t: *Tokenizer) !void {	...	@@ -797,12 +810,16 @@ fn consumeString(t: *Tokenizer) !void {
797	t.pos = esc_start;	810	t.pos = esc_start;
798	return err;	811	return err;
799	},	812	},
800	error.UnterminatedString => return err,	813	// error.UnterminatedString => return err,
801	};	814	};
802	},	815	},
803	else => return t.maybeUTF8ErrorAt(error.InvalidEscapeSequence, 1),	816	else => return t.maybeUTF8ErrorAt(error.InvalidEscapeSequence, 1),
804	}	817	}
805	},	818	},
		819	'\n' => {
		820	t.pos = start;
		821	return error.UnterminatedString;
		822	},
806	else => {	823	else => {
807	t.pos += 1;	824	t.pos += 1;
808	},	825	},
...	@@ -818,16 +835,16 @@ fn consumeStringUnicodeEscape(t: *Tokenizer) !void {	...	@@ -818,16 +835,16 @@ fn consumeStringUnicodeEscape(t: *Tokenizer) !void {
818	const trace = T.trace(@src());	835	const trace = T.trace(@src());
819	defer trace.end();	836	defer trace.end();
820		837
821	assert(t.byte().? == '\\');	838	assert(t.byte() == '\\');
822	assert(t.byteAt(1).? == 'u');	839	assert(t.nextByte() == 'u');
823	t.pos += 2;	840	t.pos += 2;
824		841
825	if ((t.byte() orelse return error.UnterminatedString) != '{')	842	if (t.byte() != '{')
826	return error.InvalidEscapeSequence;	843	return error.InvalidEscapeSequence;
827	t.pos += 1;	844	t.pos += 1;
828		845
829	if ((t.byte() orelse return error.UnterminatedString) == '0') {	846	if (t.byte() == '0') {
830	if ((t.byteAt(1) orelse return error.UnterminatedString) == 'x') {	847	if (t.nextByte() == 'x') {
831	t.pos += 2;	848	t.pos += 2;
832	}	849	}
833	}	850	}
...	@@ -837,7 +854,7 @@ fn consumeStringUnicodeEscape(t: *Tokenizer) !void {	...	@@ -837,7 +854,7 @@ fn consumeStringUnicodeEscape(t: *Tokenizer) !void {
837		854
838	const slice = t.code[start..t.pos];	855	const slice = t.code[start..t.pos];
839		856
840	if ((t.byte() orelse return error.UnterminatedString) != '}')	857	if (t.byte() != '}')
841	return error.InvalidEscapeSequence;	858	return error.InvalidEscapeSequence;
842		859
843	const code_point = std.fmt.parseInt(u21, slice, 16) catch \|err\| switch (err) {	860	const code_point = std.fmt.parseInt(u21, slice, 16) catch \|err\| switch (err) {
...	@@ -866,7 +883,7 @@ fn maybeUTF8Error(t: *Tokenizer, comptime real_err: anytype) @Type(.{	...	@@ -866,7 +883,7 @@ fn maybeUTF8Error(t: *Tokenizer, comptime real_err: anytype) @Type(.{
866	.{ .name = @errorName(real_err) },	883	.{ .name = @errorName(real_err) },
867	},	884	},
868	}) {	885	}) {
869	switch (t.byte() orelse return real_err) {	886	switch (t.byte()) {
870	0x80...0xFF => {	887	0x80...0xFF => {
871	_ = try t.codePoint();	888	_ = try t.codePoint();
872	return real_err;	889	return real_err;
...	@@ -881,7 +898,7 @@ fn maybeUTF8ErrorAt(t: *Tokenizer, comptime real_err: anytype, at: u32) @Type(.{	...	@@ -881,7 +898,7 @@ fn maybeUTF8ErrorAt(t: *Tokenizer, comptime real_err: anytype, at: u32) @Type(.{
881	.{ .name = @errorName(real_err) },	898	.{ .name = @errorName(real_err) },
882	},	899	},
883	}) {	900	}) {
884	switch (t.byteAt(at) orelse return real_err) {	901	switch (t.byteAtRt(at) orelse return real_err) {
885	0x80...0xFF => {	902	0x80...0xFF => {
886	_ = try t.codePoint();	903	_ = try t.codePoint();
887	return real_err;	904	return real_err;
...	@@ -895,7 +912,7 @@ pub fn skipWhitespace(t: *Tokenizer) void {	...	@@ -895,7 +912,7 @@ pub fn skipWhitespace(t: *Tokenizer) void {
895	defer trace.end();	912	defer trace.end();
896		913
897	while (true) {	914	while (true) {
898	switch (t.byte() orelse break) {	915	switch (t.byte()) {
899	' ', '\t', '\n', '\r' => {	916	' ', '\t', '\n', '\r' => {
900	t.pos += 1;	917	t.pos += 1;
901	},	918	},

src/compiler/Transform.zig+95-28

 @@ -14,29 +14,51 @@ scratch: std.ArrayListUnmanaged(u32) = .{},
 call_args: std.ArrayListUnmanaged(Inst.Ref) = .{},
 scopes: std.ArrayListUnmanaged(Scope) = .{},
-pub fn transform(ast: Ast, gpa: Allocator) !Teg {
+report: Report.Builder,
+pub const Error = error{
+    OutOfMemory,
+    TransformFail,
+};
+pub fn transform(
+    ast: Ast,
+    gpa: Allocator,
+    file: *Comp.File,
+    fail_report: *Report,
+) Error!Teg {
     const trace = tracer.trace(@src());
     defer trace.end();
     var t: Transform = .{
         .ast = ast,
         .gpa = gpa,
+        .report = .{ .gpa = gpa, .current_file = file },
     };
     errdefer t.deinit();
     try t.string_bytes.append(t.gpa, 0);
-    const module = try t.transformModule(0);
+    const module = t.transformModule(0) catch |err| switch (err) {
+        error.OutOfMemory => return err,
+    };
+    const report = try t.report.finalize();
+    if (report.errors.len > 0) {
+        fail_report.* = report;
+        return error.TransformFail;
+    }
     assert(@intFromEnum(module) == 0);
     try t.insts.setCapacity(t.gpa, t.insts.len);
     defer t.deinit();
     return Teg{
         .insts = t.insts.toOwnedSlice(),
         .string_bytes = try t.string_bytes.toOwnedSlice(t.gpa),
         .extra = try t.extra.toOwnedStore(t.gpa),
+        .report = report,
     };
 }
 @@ -45,6 +67,9 @@ pub fn deinit(t: *Transform) void {
     t.call_args.deinit(t.gpa);
     assert(t.scopes.items.len == 0);
     t.scopes.deinit(t.gpa);
+    t.string_bytes.deinit(t.gpa);
+    t.insts.deinit(t.gpa);
+    t.extra.deinit(t.gpa);
     t.* = undefined;
 }
 @@ -65,6 +90,25 @@ const Scope = struct {
         };
     };
+    fn addUnusedWarnings(s: Scope, t: *Transform) !void {
+        for (s.symbols.values()) |value| {
+            if (!value.used) {
+                try t.report.warn(.unused, .{
+                    .node = value.node,
+                    .kind = switch (value.data) {
+                        .declaration => .function,
+                        .param => .param,
+                        .mutable => .variable,
+                        .immutable => switch (t.ast.nodeTag(value.node)) {
+                            .param_decl => .param,
+                            else => .constant,
+                        },
+                    },
+                });
+            }
+        }
+    }
     fn deinit(s: *Scope, gpa: Allocator) void {
         s.symbols.deinit(gpa);
         s.* = undefined;
 @@ -171,7 +215,7 @@ const BuiltInIds = std.ComptimeStringMap(Inst.Ref, .{
     // todo: nan/infinity?
 });
-fn transformModule(t: *Transform, module_node: NodeIndex) !Inst.Index {
+fn transformModule(t: *Transform, module_node: NodeIndex) Allocator.Error!Inst.Index {
     const trace = tracer.trace(@src());
     defer trace.end();
 @@ -259,10 +303,12 @@ fn transformModule(t: *Transform, module_node: NodeIndex) !Inst.Index {
         .extra_index = try t.extra.append(t.gpa, IndexList, declarations),
     });
+    try t.scope().addUnusedWarnings(t);
     return module;
 }
-fn previsitModuleStatement(t: *Transform, node: NodeIndex) !void {
+fn previsitModuleStatement(t: *Transform, node: NodeIndex) Allocator.Error!void {
     const trace = tracer.trace(@src());
     defer trace.end();
 @@ -272,7 +318,11 @@ fn previsitModuleStatement(t: *Transform, node: NodeIndex) !void {
             const fn_decl = t.ast.fullFnDecl(node);
             const gop = try t.scope().symbols.getOrPut(t.gpa, fn_decl.id);
             if (gop.found_existing) {
-                return error.DuplicateFunction;
+                try t.report.err(.duplicate_function, .{
+                    .node = node,
+                    .prev = gop.value_ptr.node,
+                });
+                return;
             }
             gop.value_ptr.* = .{
                 .node = node,
 @@ -299,7 +349,7 @@ fn transformModuleStatement(t: *Transform, decl: BlockContext, node: NodeIndex)
     }
 }
-fn transformFunction(t: *Transform, outer_decl: BlockContext, node: NodeIndex) !Inst.Index {
+fn transformFunction(t: *Transform, outer_decl: BlockContext, node: NodeIndex) Allocator.Error!Inst.Index {
     const trace = tracer.trace(@src());
     defer trace.end();
 @@ -350,6 +400,8 @@ fn transformFunction(t: *Transform, outer_decl: BlockContext, node: NodeIndex) !
         .extra_index = decl_extra,
     });
+    try t.scope().addUnusedWarnings(t);
     return func;
 }
 @@ -366,7 +418,7 @@ fn transformExpressionToBlock(
     decl: BlockContext,
     node: NodeIndex,
     comptime is_function: bool,
-) anyerror!Inst.Ref {
+) Allocator.Error!Inst.Ref {
     const trace = tracer.trace(@src());
     defer trace.end();
 @@ -461,7 +513,7 @@ fn transformExpressionToBlock(
     return ret_block.toRef();
 }
-fn transformParamDecl(t: *Transform, decl: BlockContext, param: NodeIndex) !Inst.Ref {
+fn transformParamDecl(t: *Transform, decl: BlockContext, param: NodeIndex) Allocator.Error!Inst.Ref {
     const trace = tracer.trace(@src());
     defer trace.end();
 @@ -472,7 +524,11 @@ fn transformParamDecl(t: *Transform, decl: BlockContext, param: NodeIndex) !Inst
     // can use this param's value, if known at compile time.
     const sym = try t.scope().symbols.getOrPut(t.gpa, param_decl.id);
     if (sym.found_existing) {
-        return error.DuplicateParam;
+        try t.report.err(.duplicate_param, .{
+            .node = param,
+            .prev = sym.value_ptr.node,
+        });
+        return .void_value;
     }
     sym.value_ptr.* = .{
 @@ -501,7 +557,8 @@ const Builtin = enum {
         defer trace.end();
         if (args.len != 1) {
-            return error.InvalidBuiltin;
+            // try t.report.err(.invalid_builtin_args, .{ .id = "js" });
+            return .void_value;
         }
         const source = try t.transformExpression(decl, args[0]) orelse return null;
 @@ -518,7 +575,7 @@ const Builtin = enum {
 /// This is the primary transformation function: converting an AST expression into instructions.
 /// `null` represents a `return` or `break`, aka this branch of code is dead.
-fn transformExpression(t: *Transform, decl: BlockContext, node: NodeIndex) anyerror!?Inst.Ref {
+fn transformExpression(t: *Transform, decl: BlockContext, node: NodeIndex) Allocator.Error!?Inst.Ref {
     const node_tag = t.ast.nodes.items(.tag)[node];
     const trace = tracer.traceFmt(@src(), "{s}", .{@tagName(node_tag)});
 @@ -537,17 +594,13 @@ fn transformExpression(t: *Transform, decl: BlockContext, node: NodeIndex) anyer
                 return ref;
             }
-            if (mem.eql(u8, id, "_")) {
-                return error.InvalidDiscard;
-            }
             if (mem.startsWith(u8, id, "@")) {
                 @panic("TODO: string identifiers");
             }
             const lookup = t.lookupId(id) orelse {
-                std.debug.print("lookupId failed for {s}\n", .{id});
+                try t.report.err(.unknown_identifier, .{ .node = node });
-                return error.UnknownIdentifier;
+                return .void_value;
             };
             return switch (lookup.data) {
 @@ -706,7 +759,7 @@ fn transformExpression(t: *Transform, decl: BlockContext, node: NodeIndex) anyer
         .label => {
             const expr = t.ast.nodeData(node).label.expr;
-            const label = t.ast.tokenSlice(t.ast.nodes.items(.main_token)[node]);
+            const label = t.ast.tokenSlice(t.ast.nodeToken(node));
             var target: BlockContext.BreakTarget = .{
                 .parent = decl.break_target,
 @@ -778,7 +831,8 @@ fn transformExpression(t: *Transform, decl: BlockContext, node: NodeIndex) anyer
         .builtin_call,
         => {
             const builtin = Builtin.map.get(t.ast.basicNodeSlice(node)[1..]) orelse {
-                return error.InvalidBuiltin;
+                try t.report.err(.invalid_builtin, .{ .node = node });
+                return .void_value;
             };
             const base = t.ast.nodeData(node).list;
 @@ -847,7 +901,7 @@ fn transformExpression(t: *Transform, decl: BlockContext, node: NodeIndex) anyer
             const string_ref = try t.addInstRef(.{
                 .tag = .string,
                 .data = EncodedData.encode(.string, .{
-                    .src_node = decl.nodeOffset(node), // TODO: this is incorrect, there is no node
+                    .src_node = decl.nodeOffset(node),
                     .extra_index = try t.extra.append(t.gpa, SlicedString, prop_name_str),
                 }),
             });
 @@ -915,7 +969,8 @@ fn transformExpression(t: *Transform, decl: BlockContext, node: NodeIndex) anyer
             const ret_node = t.ast.nodeData(node).@"return";
             const ret_block_ptr = decl.return_target orelse {
-                return error.InvalidReturn;
+                // return error.InvalidReturn;
+                @panic("TODO");
             };
             const ret_block = if (ret_block_ptr.*) |block|
 @@ -957,7 +1012,8 @@ fn transformExpression(t: *Transform, decl: BlockContext, node: NodeIndex) anyer
             );
             const block = try decl.findAndMarkBreak(t, label) orelse {
-                return error.UnknownBreakTarget;
+                try t.report.err(.unresolved_label, .{ .node = node });
+                return null;
             };
             const operand: Inst.Ref = if (brk_node.value.unwrap()) |expr|
 @@ -1001,6 +1057,17 @@ fn transformBlockStatement(
         .var_decl => {
             const var_decl = t.ast.fullVarDecl(node);
+            if (mem.eql(u8, var_decl.id, "_")) {
+                try t.report.err(.invalid_discard, .{ .node = node });
+                return .void_value;
+            }
+            if (BuiltInIds.has(var_decl.id)) {
+                try t.report.err(.shadow_builtin_variable, .{ .node = node });
+                return .void_value;
+            }
             const type_ref = if (var_decl.type) |data_type|
                 try t.transformExpression(decl, data_type)
             else
 @@ -1040,7 +1107,7 @@ fn transformBlockStatement(
                 },
             });
-            return mem_ref;
+            return .void_value;
         },
         .fn_decl,
 @@ -1145,10 +1212,7 @@ fn getInstData(t: *Transform, index: Inst.Index) *EncodedData {
 }
 fn addScratch(t: *Transform, comptime T: type, value: T) !void {
-    try t.scratch.append(t.gpa, switch (T) {
+    try t.scratch.append(t.gpa, @intFromEnum(value));
-        Inst.Ref, Inst.Index => @intFromEnum(value),
-        else => @compileError("addScratch not implemented for " ++ @typeName(T)),
-    });
 }
 fn scratchToRefList(t: *Transform, scratch_top: usize) !RefList {
 @@ -1273,4 +1337,7 @@ const optionalExtraIndex = OptionalExtraIndex.init;
 const InternPool = @import("InternPool.zig");
+const Report = @import("Report.zig");
+const Comp = @import("Comp.zig");
 const tracer = @import("tracer");

src/compiler/Value.zig-1

...	@@ -32,7 +32,6 @@ pub fn asString(val: Value, ip: *InternPool) InternPool.String.WithLen {	...	@@ -32,7 +32,6 @@ pub fn asString(val: Value, ip: *InternPool) InternPool.String.WithLen {
32	}	32	}
33		33
34	pub fn @"type"(val: Value, ip: *InternPool) Value {	34	pub fn @"type"(val: Value, ip: *InternPool) Value {
35	std.debug.print("typeOfValue {any}\n", .{val.toKey(ip)});
36	const v: InternPool.Index = switch (val.toKey(ip)) {	35	const v: InternPool.Index = switch (val.toKey(ip)) {
37	.simple_type, .type_function => .type_type,	36	.simple_type, .type_function => .type_type,
38		37

src/compiler/main.zig+23-3

...	@@ -3,11 +3,15 @@ const Tokenizer = @import("Tokenizer.zig");	...	@@ -3,11 +3,15 @@ const Tokenizer = @import("Tokenizer.zig");
3	const Parse = @import("Parse.zig");	3	const Parse = @import("Parse.zig");
4	const Transform = @import("Transform.zig");	4	const Transform = @import("Transform.zig");
5	const Comp = @import("Comp.zig");	5	const Comp = @import("Comp.zig");
6		6	const unicode_props = @import("unicode_props.zig");
7	const T = @import("tracer");	7	const T = @import("tracer");
8	pub const tracer_impl = @field(T, "none");	8	pub const tracer_impl = @field(T, "none");
9		9
		10	pub const Report = @import("Report.zig");
		11
10	pub fn main() !void {	12	pub fn main() !void {
		13	try unicode_props.init();
		14
11	std.debug.print("\r\x1b[K", .{});	15	std.debug.print("\r\x1b[K", .{});
12		16
13	try T.init();	17	try T.init();
...	@@ -45,7 +49,14 @@ pub fn main() !void {	...	@@ -45,7 +49,14 @@ pub fn main() !void {
45	defer alloc.free(exe_name);	49	defer alloc.free(exe_name);
46		50
47	const file_path = maybe_file_path orelse {	51	const file_path = maybe_file_path orelse {
48	std.debug.print("Usage: {s} <input file>\n", .{exe_name});	52	std.debug.print("Usage: {s} [flags] <input file>\n", .{exe_name});
		53	std.debug.print(
		54	// \| 60 chars wrap point\|
		55	\\
		56	\\Flags:
		57	\\ --dump-ir Dump intermediate representations.
		58	\\
		59	, .{});
49	std.process.exit(1);	60	std.process.exit(1);
50	};	61	};
51		62
...	@@ -54,7 +65,16 @@ pub fn main() !void {	...	@@ -54,7 +65,16 @@ pub fn main() !void {
54	defer comp.deinit();	65	defer comp.deinit();
55	const main_file = try comp.getFile(file_path);	66	const main_file = try comp.getFile(file_path);
56		67
57	const teg = try main_file.getTeg(comp.gpa);	68	var fail_report: Report = undefined;
		69	const teg = main_file.getTeg(&comp, &fail_report) catch \|err\| switch (err) {
		70	error.TransformFail => {
		71	std.debug.print("Transform failed:\n", .{});
		72	try fail_report.dumpToStdErr(&comp);
		73	std.process.exit(1);
		74	},
		75	else => return err,
		76	};
		77
58	{	78	{
59	const stderr = std.io.getStdErr().writer();	79	const stderr = std.io.getStdErr().writer();
60	try teg.dumpToWriter(stderr);	80	try teg.dumpToWriter(stderr);

src/compiler/util.zig+2-2

...	@@ -7,9 +7,9 @@ pub fn detectTTYFromWriter(writer: anytype) std.io.tty.Config {	...	@@ -7,9 +7,9 @@ pub fn detectTTYFromWriter(writer: anytype) std.io.tty.Config {
7	return std.io.tty.detectConfig(writer.context);	7	return std.io.tty.detectConfig(writer.context);
8	}	8	}
9	if (@TypeOf(writer) == std.io.AnyWriter) {	9	if (@TypeOf(writer) == std.io.AnyWriter) {
10	const writeFn = comptime (std.fs.File{ .handle = undefined })	10	const fileWriteFn = comptime (std.fs.File{ .handle = undefined })
11	.writer().any().writeFn;	11	.writer().any().writeFn;
12	if (writer.writeFn == writeFn) {	12	if (writer.writeFn == fileWriteFn) {
13	const file: *const std.fs.File = @alignCast(@ptrCast(writer.context));	13	const file: *const std.fs.File = @alignCast(@ptrCast(writer.context));
14	return std.io.tty.detectConfig(file.*);	14	return std.io.tty.detectConfig(file.*);
15	}	15	}

author	chloe caruso <git@paperclover.net> 2024-05-19 03:11:59-07:00
committer	chloe caruso <git@paperclover.net> 2024-05-19 03:11:59-07:00
log	cf749893b592c342cd21fe524c7c45456876cae3
tree	10775f14eb33a1b28aecbc4558bb5214ed732b94
parent	1b47741300f878f330f2efca4141a92e91945b78

...	@@ -19,7 +19,6 @@ pub fn linkJsCode(	...	@@ -19,7 +19,6 @@ pub fn linkJsCode(
19	decls: []const Decl.Index,	19	decls: []const Decl.Index,
20	writer: anytype,	20	writer: anytype,
21	) !void {	21	) !void {
22	log.info("Linking {d} decls", .{decls.len});
23		22
24	// Arena allocator is used for the main flow	23	// Arena allocator is used for the main flow
25	var link: JsLink = .{ .comp = comp };	24	var link: JsLink = .{ .comp = comp };
...	@@ -68,7 +67,7 @@ pub fn linkJsCode(	...	@@ -68,7 +67,7 @@ pub fn linkJsCode(
68	try link.decls.putNoClobber(comp.gpa, decl_index, @enumFromInt(gop.index));	67	try link.decls.putNoClobber(comp.gpa, decl_index, @enumFromInt(gop.index));
69	}	68	}
70		69
71	try writer.print("// Compiler Demo. Do not use in production\n", .{});	70	try writer.print("// Language Compiler v0.0.1-alpha2\n", .{});
72		71
73	// 2. Print every function decl	72	// 2. Print every function decl
74	for (decls[1..]) \|decl_index\| {	73	for (decls[1..]) \|decl_index\| {
...	@@ -106,7 +105,6 @@ fn reserveExternSymbol(link: *JsLink, extern_symbol: String.WithLen) !void {	...	@@ -106,7 +105,6 @@ fn reserveExternSymbol(link: *JsLink, extern_symbol: String.WithLen) !void {
106		105
107	fn appendLinkString(link: *JsLink, string: []const u8) !LinkString {	106	fn appendLinkString(link: *JsLink, string: []const u8) !LinkString {
108	const start = link.string_bytes.items.len;	107	const start = link.string_bytes.items.len;
109	log.info("Append '{s}' -> .{{ .start = {d}, .len = {d} }}", .{ string, start, string.len });
110	try link.string_bytes.appendSlice(link.comp.gpa, string);	108	try link.string_bytes.appendSlice(link.comp.gpa, string);
111	return .{ .start = @intCast(start), .len = @intCast(string.len) };	109	return .{ .start = @intCast(start), .len = @intCast(string.len) };
112	}	110	}
...	@@ -170,7 +168,6 @@ fn printBlock(link: JsLink, tree: const JsTree, list: JsTree.IndexList, writer	...	@@ -170,7 +168,6 @@ fn printBlock(link: JsLink, tree: const JsTree, list: JsTree.IndexList, writer
170	}	168	}
171		169
172	fn printExpr(link: JsLink, tree: const JsTree, node: Node.Index, writer: anytype) @TypeOf(writer).Error!void {	170	fn printExpr(link: JsLink, tree: const JsTree, node: Node.Index, writer: anytype) @TypeOf(writer).Error!void {
173	log.info("print node {d}", .{@intFromEnum(node)});
174	switch (tree.nodeTag(node)) {	171	switch (tree.nodeTag(node)) {
175	else => \|tag\| std.debug.panic("TODO: printExpr .{s}", .{@tagName(tag)}),	172	else => \|tag\| std.debug.panic("TODO: printExpr .{s}", .{@tagName(tag)}),
176	.root => unreachable,	173	.root => unreachable,

better error handling