From 5b5a3038aa08c430f60533153b0823267b3dbdff Mon Sep 17 00:00:00 2001
From: chloe caruso <git@paperclover.net>
Date: Sun, 21 Jul 2024 15:39:07 -0700
Subject: [PATCH] remove $ as a valid identifier character

The casing conventions have not been set, but there is no universe where
we allow $ as a recommended identifier character. we had this initially
because JavaScript allows it, but in practice people just use it in
frameworks to mean "magic". (see jQuery, Svelte, Qwik, etc)
---
 CHANGELOG.md           |  5 +++--
 src/core/Report.zig    |  7 +------
 src/core/Tokenizer.zig |  6 +++---
 src/core/Transform.zig | 20 ++++++++------------
 4 files changed, 15 insertions(+), 23 deletions(-)

diff --git a/CHANGELOG.md b/CHANGELOG.md
index 155e6db9170c549615b492f72d313e96f5247013..078597fe175ba66eb41bb8f117258af5117ee87b 100644
--- a/CHANGELOG.md
+++ b/CHANGELOG.md
@@ -1,9 +1,10 @@
 ## 0.1.4
 
-- feat: implement string identifiers
-- feat: implement basic struct types
+- feat: implement basic struct types (#42)
+- feat: implement string identifiers (#54)
 - feat: include a `README.html` alongside compiler with documentation and code samples.
 - feat(transform): include a warning for leading zero in numbers
+- feat(tokenizer): `$` is no longer valid in identifiers
 - fix(sema): miscompilation using left side of all binary expression for both sides
 - fix(sema): handle recursive function calls
 - fix(sema): quit block analysis when hitting `noreturn`
diff --git a/src/core/Report.zig b/src/core/Report.zig
index 49bb6ebd1250af3acd43aa534357a5d3cd4d3197..317ee696d37a439c83265b4c73c23757d2fa3c71 100644
--- a/src/core/Report.zig
+++ b/src/core/Report.zig
@@ -1130,12 +1130,7 @@ fn formatIdImpl(
         var i: usize = 0;
         while (i < inner.len) {
             switch (inner[i]) {
-                'a'...'z',
-                'A'...'Z',
-                '0'...'9',
-                '_',
-                '$',
-                => {
+                'a'...'z', 'A'...'Z', '0'...'9', '_' => {
                     i += 1;
                     continue;
                 },
diff --git a/src/core/Tokenizer.zig b/src/core/Tokenizer.zig
index d8583b1b8e274716d41c8ab53ddaf14e725e2d29..910ecb2ae6bdb8eab88975580815f139bdf6e79e 100644
--- a/src/core/Tokenizer.zig
+++ b/src/core/Tokenizer.zig
@@ -322,7 +322,7 @@ pub fn next(t: *Tokenizer) !?Token {
         switch (t.byte()) {
             ' ', '\t', '\n', '\r' => unreachable, // call skipWhitespace
 
-            'a'...'z', 'A'...'Z', '_', '$' => {
+            'a'...'z', 'A'...'Z', '_' => {
                 const start = t.pos;
                 t.pos += 1;
                 return try t.tokenizeIdentifier(start);
@@ -563,7 +563,7 @@ fn codePoint(t: Tokenizer) !struct { u3, u21 } {
 
 fn isAtIdentifierStart(t: *Tokenizer) !bool {
     switch (t.byte()) {
-        'a'...'z', 'A'...'Z', '_', '$' => return true,
+        'a'...'z', 'A'...'Z', '_' => return true,
         else => return false,
         0x80...0xFF => {
             _, const code_point = try t.codePoint();
@@ -578,7 +578,7 @@ fn tokenizeIdentifier(t: *Tokenizer, start_byte: u32) !Token {
 
     while (true) {
         switch (t.byte()) {
-            'a'...'z', 'A'...'Z', '0'...'9', '_', '$' => {
+            'a'...'z', 'A'...'Z', '0'...'9', '_' => {
                 t.pos += 1;
             },
             0x80...0xFF => {
diff --git a/src/core/Transform.zig b/src/core/Transform.zig
index e0013ca2c468039ed62ee36581986f8920dc52cb..2cf3d708e317aba9ee69c2d1b3811421b3e0cb00 100644
--- a/src/core/Transform.zig
+++ b/src/core/Transform.zig
@@ -6,8 +6,8 @@
 //! - Annotating expression statements with `.discard` and `.ensure_void`
 //! - Flattening nested expressions into instruction sequences.
 //!
-//! Some trivial diagnostics such as unused variables, unknown result locations, and
-//! invalid assignment targets are also reported here.
+//! Some trivial diagnostics such as unused variables, unknown result types,
+//! invalid assignment targets, and other style issues are also reported here.
 gpa: Allocator,
 ast: Ast,
 insts: Inst.List = .{},
@@ -22,9 +22,9 @@ scopes: std.ArrayListUnmanaged(Scope) = .{},
 report: Report.Builder,
 
 pub const Error = error{
-    OutOfMemory,
+    /// A `Report` will contain at least one error.
     TransformFail,
-};
+} || Allocator.Error;
 
 pub fn transform(
     ast: Ast,
@@ -61,7 +61,7 @@ pub fn transform(
 
     try t.insts.setCapacity(t.gpa, t.insts.len);
 
-    return Teg{
+    return .{
         .insts = t.insts.toOwnedSlice(),
         .string_bytes = try t.string_bytes.toOwnedSlice(t.gpa),
         .extra = try t.extra.toOwnedStore(t.gpa),
@@ -142,18 +142,14 @@ const Scope = struct {
     }
 };
 
-const ScratchType = enum {
-    ref,
-    index,
-};
-
 const BlockContext = struct {
     node_base: i32,
     return_target: ?*?Inst.Index = null,
     break_target: ?*BreakTarget = null,
 
-    /// This struct is always stored on the stack; the idea is to pass
-    /// mutable pointers to it. When `break :brk` is encountered
+    /// This struct is always stored on the stack; the idea is to pass mutable
+    /// pointers to it. When `break :brk` is encountered, it will generate the
+    /// block instruction. This way, redundant blocks are never created.
     const BreakTarget = struct {
         label: ?[]const u8,
         /// This allows `continue` as well as implicit break targeting.
-- 
2.54.0