Commit 227021403e3604e6fc049ad2fa7575b05548244b | ix

author	Kai Tamkun <13513421+heimskr@users.noreply.github.com> 2024-06-02 21:32:46-07:00
committer	GitHub <noreply@github.com> 2024-06-02 21:32:46-07:00
log	227021403e3604e6fc049ad2fa7575b05548244b
tree	06836771054a46299091cc2111735cc17a25728a
parent	f499572ffea4cab93d28e2480bcc9a4234f59089

chore: fix codebase and manual typos (#35)

12 files changed, 49 insertions(+), 49 deletions(-)

doc
- language-manual.md
src
- core

doc/language-manual.md+6-6

...	...	@@ -11,8 +11,8 @@ Key goals of this project are:
11	11	It primarily focuses on merging some great points of Rust, Zig, and JavaScript's language designs:
12	12
13	13	- Rust - Expressive control flow that is easy to write.
14		- Zig - Execute any code at compile time, everything is an expression.
15		- JavaScript - Garbage collected values with a versitile type system.
	14	- Zig - Execute any code at compile time; everything is an expression.
	15	- JavaScript - Garbage collected values with a versatile type system.
16	16
17	17	## Standard Library
18	18
...	...	@@ -25,9 +25,9 @@ let print = @js("console").log;
25	25	print("Hello, world!");
26	26	```
27	27
28		The `print` function is implemented by getting an external reference to the `console` JavaScript global, and then retriving the `log` property from it.
	28	The `print` function is implemented by getting an external reference to the `console` JavaScript global and then retrieving the `log` property from it.
29	29
30		## Primative Data Types
	30	## Primitive Data Types
31	31
32	32	\| name \| ts equivilent \| description \|
33	33	\| ---------- \| --------------------- \| ---------------------------------------------- \|
...	...	@@ -36,7 +36,7 @@ The `print` function is implemented by getting an external reference to the `con
36	36	\| `bool` \| `boolean` \| either `true` or `false` \|
37	37	\| `string` \| `string` \| a sequence of text characters \|
38	38	\| `void` \| `undefined` or `null` \| detectable lack of a value \|
39		\| `opaque` \| `unknown` or `any` \| could be any value \|
	39	\| `opaque` \| `unknown` or `any` \| could be any value \|
40	40	\| `noreturn` \| `never` \| control flow prevents this value from existing \|
41	41
42	42	> note: `bigint` is planned but not added
...	...	@@ -61,7 +61,7 @@ let true = 1 == 0;
61	61
62	62	There are two types of variables, `let` and `var`.
63	63
64		`let` bindings cannot be overridden, and should be thought of as aliases of the expression within. Referring to these identifiers is equivilent to inlining it's where it was used, aside from side effect duplication.
	64	`let` bindings cannot be overridden, and should be thought of as aliases of the expression within. Referring to these identifiers is equivalent to inlining it where it was used, aside from side effect duplication.
65	65
66	66	```lang:run
67	67	let hello: int = 1;

src/core/Ast.zig+6-6

...	...	@@ -1,11 +1,11 @@
1	1	//! The Abstract Syntax Tree preserves most details of the source code,
2	2	//! as it is used for formatting. Before semantic analysis, this
3	3	//! structure is normalized into `Teg`, which removes constructs like
4		//! grouping, unneded let bindings, and other transforms.
	4	//! grouping, unneeded let bindings, and other transforms.
5	5	//!
6	6	//! `ast.nodes` contains nodes stored as three parts:
7	7	//! - tag: the type of the node
8		//! - main_token: index to it's primary token, depends on the tag
	8	//! - main_token: index to its primary token, depends on the tag
9	9	//! - data: two u32 values, depends on the tag
10	10	//!
11	11	//! Many nodes encode all their metadata with just `node.data`, using
...	...	@@ -20,7 +20,7 @@
20	20	//! - If the keyword before `main_token` is `kw_pub` it is a public variable.
21	21	//!
22	22	//! To make decoding these easier, many nodes contain a "full" variant,
23		//! such as `full.VarDecl`, which can be aquired with `ast.fullVarDecl(id)`.
	23	//! such as `full.VarDecl`, which can be acquired with `ast.fullVarDecl(id)`.
24	24	//!
25	25	//! For nodes that must store more than 64 bits of payload data, an "extra"
26	26	//! struct is stored out-of-line in the `extra` field. These can be read
...	...	@@ -56,7 +56,7 @@ pub const Node = struct {
56	56	/// main_token is the {
57	57	/// data is .list (containing stmt slice)
58	58	///
59		/// This guarantees that the block either empty, or the last statement
	59	/// This guarantees that either the block is empty or the last statement
60	60	/// ends with a semicolon.
61	61	block_void,
62	62	/// main_token is the {
...	...	@@ -332,7 +332,7 @@ pub const Node = struct {
332	332	extra: ExtraIndex,
333	333	body: NodeIndex,
334	334	},
335		/// either is a node list or a token list, depending on the node's tag
	335	/// is either a node list or a token list, depending on the node's tag
336	336	list: SubList,
337	337	var_decl: struct {
338	338	type: OptionalNodeIndex,
...	...	@@ -398,7 +398,7 @@ pub const Span = struct {
398	398	main: u32,
399	399	};
400	400
401		// This can be removed when the following issue is implemented:
	401	// This can be removed when the following issues are implemented:
402	402	// See: https://github.com/ziglang/zig/issues/104
403	403	// See: https://github.com/ziglang/zig/issues/3806
404	404	pub const OptionalNodeIndex = enum(u32) {

src/core/Cache.zig+1-1

...	...	@@ -1,6 +1,6 @@
1	1	const std = @import("std");
2	2
3		/// This is 128 bits - Even with 2^54 cache entries, the probably of a collision would be under 10^-6
	3	/// This is 128 bits - Even with 2^54 cache entries, the probability of a collision would be under 10^-6
4	4	pub const bin_digest_len = 16;
5	5	pub const hex_digest_len = bin_digest_len * 2;
6	6	pub const BinDigest = [bin_digest_len]u8;

src/core/Comp.zig+6-6

...	...	@@ -12,7 +12,7 @@ analyzed_instructions: usize = 0,
12	12
13	13	options: Options = .{},
14	14
15		// TODO: consider computing this dynamicly per-link
	15	// TODO: consider computing this dynamically per-link
16	16	current_compile_decls: std.ArrayListUnmanaged(Decl.Index) = .{},
17	17
18	18	pub fn init(comp: *Comp) !void {
...	...	@@ -60,7 +60,7 @@ pub const File = struct {
60	60	root_module: Decl.Index.Optional,
61	61	status: enum {
62	62	never_loader,
63		/// A failure occurred during loading the file. Retrying might succeed.
	63	/// A failure occurred while loading the file. Retrying might succeed.
64	64	retryable_failure,
65	65	/// Errors during parse
66	66	parse_failure,
...	...	@@ -232,7 +232,7 @@ pub const Decl = struct {
232	232	teg_decl_index: InternPool.TrackedInst.Index.Optional,
233	233
234	234	/// Represents the "shallow" analysis status. For example, for decls that are functions,
235		/// the function type is analyzed with this set to `in_progress`, however, the semantic
	235	/// the function type is analyzed with this set to `in_progress`; however, the semantic
236	236	/// analysis of the function body is performed with this value set to `success`. Functions
237	237	/// have their own analysis status field.
238	238	analysis: enum {
...	...	@@ -256,10 +256,10 @@ pub const Decl = struct {
256	256	sema_failure,
257	257	/// There will be a corresponding ErrorMsg in Zcu.failed_decls.
258	258	codegen_failure,
259		/// Sematic analysis and constant value codegen of this Decl has
	259	/// Semantic analysis and constant value codegen of this Decl has
260	260	/// succeeded. However, the Decl may be outdated due to an in-progress
261	261	/// update. Note that for a function, this does not mean codegen of the
262		/// function body succeded: that state is indicated by the function's
	262	/// function body succeeded: that state is indicated by the function's
263	263	/// `analysis` field.
264	264	complete,
265	265	},
...	...	@@ -268,7 +268,7 @@ pub const Decl = struct {
268	268
269	269	is_pub: bool,
270	270
271		/// What kind of a declaration is this.
	271	/// What kind of a declaration this is.
272	272	kind: Kind,
273	273
274	274	pub fn deinit(decl: *Decl, gpa: Allocator) void {

src/core/InternPool.zig+4-4

...	...	@@ -8,7 +8,7 @@ items: std.MultiArrayList(Item) = .{},
8	8	extra: ExtraStore = .{},
9	9
10	10	/// In order to store references to strings in fewer bytes, we copy all
11		/// string bytes into here. String bytes can be null. It is up to whomever
	11	/// string bytes into here. String bytes can be null. It is up to whoever
12	12	/// is referencing the data here whether they want to store both index and length,
13	13	/// thus allowing null bytes, or store only index, and use null-termination. The
14	14	/// `string_bytes` array is agnostic to either usage.
...	...	@@ -69,7 +69,7 @@ pub const EncodedData = dod.EncodedData(Tag, u32);
69	69	pub const ExtraIndex = ExtraStore.Index;
70	70
71	71	pub const Tag = enum(u8) {
72		/// This special tag represents a value which was removed from this pool via
	72	/// This special tag represents a value that was removed from this pool via
73	73	/// `InternPool.remove`. The item remains allocated to preserve indices, but
74	74	/// lookups will consider it not equal to any other item, and all queries
75	75	/// assert not this tag. `data` is unused.
...	...	@@ -581,7 +581,7 @@ pub const TrackedInst = extern struct {
581	581	inst: Teg.Inst.Index,
582	582
583	583	comptime {
584		// The fields should be tightly packed. See also serialiation logic in `Compilation.saveState`.
	584	// The fields should be tightly packed. See also serialization logic in `Compilation.saveState`.
585	585	assert(@sizeOf(@This()) == Cache.bin_digest_len + @sizeOf(Teg.Inst.Index));
586	586	}
587	587
...	...	@@ -882,7 +882,7 @@ pub fn init(ip: *InternPool, gpa: Allocator) !void {
882	882	try ip.map.ensureUnusedCapacity(gpa, static_keys.len);
883	883	try ip.extra.list.ensureUnusedCapacity(gpa, static_keys.len);
884	884
885		// This inserts all the statically-known values into the intern pool in the
	885	// This inserts all the statically known values into the intern pool in the
886	886	// order expected.
887	887	for (static_keys) \|key\| {
888	888	_ = ip.get(gpa, key) catch unreachable;

src/core/JsLink.zig+2-2

...	...	@@ -1,5 +1,5 @@
1	1	/// This is the JavaScript backend's linker.
2		/// It takes in a compilation, a list of `JsTree`s and creates a JavaScript
	2	/// It takes in a compilation and a list of `JsTree`s and creates a JavaScript
3	3	/// bundle out of them.
4	4	comp: *Comp,
5	5
...	...	@@ -453,7 +453,7 @@ pub fn indexOfNeedsEscape(slice: []const u8) ?usize {
453	453	return null;
454	454	}
455	455
456		// This function is loosely based off of Bun's `writeJSONString`, which
	456	// This function is loosely based on Bun's `writeJSONString`, which
457	457	// is a Zig port of esbuild's `printUnquotedUTF16` function.
458	458	// Both projects are licensed under MIT
459	459	fn writeString(decoded: []const u8, writer: anytype) !void {

src/core/Parse.zig+5-5

...	...	@@ -1,5 +1,5 @@
1	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, but transferring into an `Ast` does move token list ownership.
3	3	gpa: Allocator,
4	4	source: [:0]const u8,
5	5	token_tags: []const Token.Tag,
...	...	@@ -250,7 +250,7 @@ fn parseModuleMembers(p: *Parse) Error!SubList {
250	250	try p.scratch.append(p.gpa, try p.parseExpression(true));
251	251
252	252	switch (first_tag) {
253		// some control flow and blocks are allowed to not have a
	253	// some control flow and blocks are allowed to not have
254	254	// trailing semicolons. this is a stylistic choice.
255	255	.kw_if,
256	256	.start_brace,
...	...	@@ -425,7 +425,7 @@ fn parseBlockContents(p: *Parse) Error!struct { Node.Tag, SubList } {
425	425	try p.scratch.append(p.gpa, try p.parseExpression(true));
426	426
427	427	switch (first_tag) {
428		// some control flow and blocks are allowed to not have a
	428	// some control flow and blocks are allowed to not have
429	429	// trailing semicolons. this is a stylistic choice.
430	430	.kw_if,
431	431	.start_brace,
...	...	@@ -1027,13 +1027,13 @@ fn parseGroup(p: *Parse) Error!NodeIndex {
1027	1027
1028	1028	// helpers
1029	1029
1030		/// Append a node and return it's index.
	1030	/// Append a node and return its index.
1031	1031	fn addNode(p: *Parse, node: Node) !NodeIndex {
1032	1032	const trace = t.trace(@src());
1033	1033	defer trace.end();
1034	1034
1035	1035	try p.nodes.append(p.gpa, node);
1036		// @intCast is always safe because 2^32 tokens alwayss produce less than
	1036	// @intCast is always safe because 2^32 tokens always produce less than
1037	1037	// 2^32 nodes, and there is a 2^32 limit on source file bytes.
1038	1038	return @intCast(p.nodes.len - 1);
1039	1039	}

src/core/Report.zig+3-3

...	...	@@ -358,7 +358,7 @@ pub const Warn = union(enum) {
358	358	},
359	359	};
360	360
361		/// A variant of std.builtin.StackTrace that stores it's data within itself
	361	/// A variant of std.builtin.StackTrace that stores its data within itself
362	362	/// instead of being a pointer. This allows capturing the stack trace and
363	363	/// saving it for later printing.
364	364	pub const StoredTrace = struct {
...	...	@@ -603,7 +603,7 @@ pub fn dumpToStdErr(report: Report, comp: *Comp) !void {
603	603	const tty = std.io.tty.detectConfig(stderr);
604	604
605	605	if (false or
606		// Windows API uses it's own API calls, so a buffered writer
	606	// Windows API uses its own API calls, so a buffered writer
607	607	// cannot be used :/
608	608	(builtin.os.tag == .windows and tty == .windows_api) or
609	609	// Debug stack trace printer writes directly to standard err,
...	...	@@ -776,7 +776,7 @@ pub const MsgContext = struct {
776	776	report: *const Report,
777	777	utf8_source: [:0]const u8,
778	778
779		/// This arena is is a stack fallback with 16kb, then
	779	/// This arena is a stack fallback with 16kb, then
780	780	/// falling back to an arena allocator. This should
781	781	/// make the cost of arbitrarily creating `Component`
782	782	/// nodes extremely lightweight.

src/core/Sema.zig+1-1

...	...	@@ -42,7 +42,7 @@ pub const Error = error{
42	42	} \|\| Comp.File.GetTegError;
43	43
44	44	/// Instead of using a hash map for this, just pre-allocate a list long enough to hold everything.
45		/// This is fine because all instructions are going to be adacent in a single declaration.
	45	/// This is fine because all instructions are going to be adjacent in a single declaration.
46	46	const InstMap = struct {
47	47	items: []Air.Inst.Ref.Optional,
48	48	start: Teg.Inst.Index,

src/core/Teg.zig+2-2

...	...	@@ -207,7 +207,7 @@ pub const Inst = struct {
207	207	}
208	208	};
209	209
210		/// A reference to instruction, or to an InternPool index, or neither.
	210	/// A reference to an instruction, or to an InternPool index, or to neither.
211	211	///
212	212	/// If the integer tag value is < InternPool.static_len, then it
213	213	/// corresponds to an InternPool index. Otherwise, this refers to a ZIR
...	...	@@ -328,7 +328,7 @@ pub const Inst = struct {
328	328	};
329	329	};
330	330
331		/// Trailng data for function declarations
	331	/// Trailing data for function declarations
332	332	pub const DeclFn = struct {
333	333	params: RefList,
334	334	return_type: Ref,

src/core/Tokenizer.zig+10-10

...	...	@@ -1,10 +1,10 @@
1		//! The tokenizer is extremely simple structure that, given a view of source
	1	//! The tokenizer is an extremely simple structure that, given a view of source
2	2	//! code, can stream tokens one at a time. A token is an 8-bit tag and a 32-bit
3	3	//! offset to where that token is located.
4	4	//!
5		//! Retriving payloads like string/numeric values, identifier names, etc is
	5	//! Retrieving payloads like string/numeric values, identifier names, etc is
6	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 guarantee that such data is valid.
8	8	code: [:0]const u8,
9	9	pos: ByteOffset = 0,
10	10
...	...	@@ -146,7 +146,7 @@ pub const Token = struct {
146	146	/// Used so that parser does not need to do out of bounds checks
147	147	eof,
148	148
149		/// Most tokens have statically known lexeme, such as "struct" or "+="
	149	/// Most tokens have statically known lexemes, such as "struct" or "+="
150	150	pub fn lexeme(tag: Tag) ?[]const u8 {
151	151	return switch (tag) {
152	152	.invalid => unreachable,
...	...	@@ -290,7 +290,7 @@ pub const Token = struct {
290	290	});
291	291	};
292	292
293		/// When any of these error occurs, the tokenizer's position is set to
	293	/// When any of these errors occurs, the tokenizer's position is set to
294	294	/// the start of the offending token.
295	295	pub const NextTokenError = error{
296	296	/// Generic token syntax error
...	...	@@ -525,7 +525,7 @@ fn nextByte(t: Tokenizer) u8 {
525	525	return t.code[t.pos + 1];
526	526	}
527	527
528		/// Remember that using this only makes sense if you know for certain that all
	528	/// Remember that using this makes sense only if you know for certain that all
529	529	/// previous bytes are ASCII, otherwise you may hit UB
530	530	fn byteAt(t: Tokenizer, comptime offset: u32) ?u8 {
531	531	if (offset == 1) @compileError("use nextByte()");
...	...	@@ -940,7 +940,7 @@ fn simpleToken(t: *Tokenizer, comptime tag: Token.Tag) Token {
940	940
941	941	/// We are about to return an error, such as error.TrailingNumericSeparator.
942	942	/// However, the character we ran into might actually be invalid UTF-8.
943		/// This utility just checks for that to make sure we dont mask error.InvalidUtf8
	943	/// This utility just checks for that to make sure we don't mask error.InvalidUtf8
944	944	fn maybeUTF8Error(t: *Tokenizer, comptime real_err: anytype) @Type(.{
945	945	.ErrorSet = &[_]std.builtin.Type.Error{
946	946	.{ .name = "InvalidUtf8" },
...	...	@@ -1003,7 +1003,7 @@ pub fn tokenLocation(
1003	1003	.line_end = @intCast(utf8_source.len),
1004	1004	};
1005	1005
1006		// Scan to by line until we go past the token start
	1006	// Scan by line until we go past the token start
1007	1007	while (std.mem.indexOfScalarPos(u8, utf8_source, loc.line_start, '\n')) \|i\| {
1008	1008	if (i >= token_start) {
1009	1009	break; // Went past
...	...	@@ -1058,7 +1058,7 @@ pub fn decodeStringCount(esc: []const u8) usize {
1058	1058	// - 0x00 thru 0x7F -> one byte
1059	1059	// - 0x80 thru 0xFF -> two bytes
1060	1060	//
1061		// we check the second byte anyways to ensure safe modes
	1061	// we check the second byte anyway to ensure safe modes
1062	1062	// catch potential invalid string literals.
1063	1063	'x' => switch (esc[backslash + 3]) {
1064	1064	'0'...'9', 'a'...'f', 'A'...'F' => {
...	...	@@ -1105,7 +1105,7 @@ pub fn decodeStringCount(esc: []const u8) usize {
1105	1105	}
1106	1106
1107	1107	/// Given a valid string literal contents (excluding the quotes), write
1108		/// it's decoded form (WTF-8) into the buf.
	1108	/// its decoded form (WTF-8) into the buf.
1109	1109	pub fn decodeStringBuf(esc: []const u8, buf: []u8) []u8 {
1110	1110	var result_i: usize = 0;
1111	1111	var i: usize = 0;

src/core/Transform.zig+3-3

...	...	@@ -144,7 +144,7 @@ const BlockContext = struct {
144	144	/// mutable pointers to it. When `break :brk` is encountered
145	145	const BreakTarget = struct {
146	146	label: ?[]const u8,
147		/// This allows `continue` as well as implicit break targetting.
	147	/// This allows `continue` as well as implicit break targeting.
148	148	is_loop: bool,
149	149	/// If `null`, this label was unused.
150	150	block_index: ?Inst.Index = null,
...	...	@@ -711,7 +711,7 @@ fn transformExpression(t: *Transform, decl: BlockContext, node: NodeIndex) Alloc
711	711	}),
712	712	.mutable => \|idx\| idx.toRef(),
713	713	.immutable => \|ref\| ref,
714		.param => unreachable, //unresolved param
	714	.param => unreachable, // unresolved param
715	715	};
716	716	},
717	717
...	...	@@ -1038,7 +1038,7 @@ fn transformExpression(t: *Transform, decl: BlockContext, node: NodeIndex) Alloc
1038	1038
1039	1039	if (yes == .void and no == .void) {
1040	1040	// when both branches do nothing, we can omit the branch,
1041		// however we still need to type-check the condition operand
	1041	// but we still need to type-check the condition operand
1042	1042	_ = try t.addInstRef(.{
1043	1043	.tag = .ensure_bool,
1044	1044	.data = EncodedData.encode(.ensure_bool, .{