fix(hamt): shrink root pointer by 8 bytes

.github/workflows/main.yml

@@ -0,0 +1,40 @@
+name: Nightly
+
+on:
+  push:
+    paths:
+      - "**.zig"
+    branches:
+      - main
+  schedule:
+    - cron: "0 0 * * *"
+  workflow_dispatch:
+
+jobs:
+  lint:
+    runs-on: ubuntu-latest
+    steps:
+      - uses: goto-bus-stop/setup-zig@v2
+        with:
+          version: 0.12.0-dev.2063+804cee3b9
+      - uses: actions/checkout@v3
+      - run: zig fmt src/*.zig
+  test:
+    strategy:
+      matrix:
+        os: [ubuntu-latest, windows-latest, macos-latest]
+    runs-on: ${{matrix.os}}
+    steps:
+      - uses: goto-bus-stop/setup-zig@v2
+        with:
+          version: 0.12.0-dev.2063+804cee3b9
+      - uses: actions/checkout@v3
+      - run: zig build test
+  bench:
+    runs-on: ubuntu-latest
+    steps:
+      - uses: goto-bus-stop/setup-zig@v2
+        with:
+          version: 0.12.0-dev.2063+804cee3b9
+      - uses: actions/checkout@v3
+      - run: zig build -Doptimize=ReleaseFast -Dcpu=baseline bench

.gitignore

@@ -1,3 +1,4 @@
+.zig-cache/
 zig-cache/
 zig-out/
-.vscode/
+.vscode/

build.zig

@@ -15,13 +15,13 @@ pub fn build(b: *std.Build) void {
     // set a preferred release mode, allowing the user to decide how to optimize.
     const optimize = b.standardOptimizeOption(.{});
 
-    const hamt_module = b.addModule("hamt", .{ .source_file = .{ .path = "src/lib.zig" } });
+    const mod = b.addModule("hamt", .{ .root_source_file = b.path("src/lib.zig") });
 
     const lib = b.addStaticLibrary(.{
         .name = "hamt",
         // In this case the main source file is merely a path, however, in more
         // complicated build scripts, this could be a generated file.
-        .root_source_file = .{ .path = "src/lib.zig" },
+        .root_source_file = b.path("src/lib.zig"),
         .target = target,
         .optimize = optimize,
     });
@@ -33,30 +33,32 @@ pub fn build(b: *std.Build) void {
 
     // Benchmark
     const bench = b.addExecutable(.{
-        .name = "hamt-benchmark",
+        .name = "hamt-bench",
-        .root_source_file = .{ .path = "src/bench.zig" },
+        .root_source_file = b.path("src/bench.zig"),
         .target = target,
-        .optimize = optimize,
+        .optimize = .ReleaseFast,
+        .link_libc = true,
     });
-    bench.addModule("hamt", hamt_module);
+    bench.root_module.addImport("hamt", mod);
 
     const bench_cmd = b.addRunArtifact(bench);
+
     const bench_step = b.step("bench", "Run benchmark");
     bench_step.dependOn(&bench_cmd.step);
 
     // Creates a step for unit testing. This only builds the test executable
     // but does not run it.
-    const main_tests = b.addTest(.{
+    const lib_unit_tests = b.addTest(.{
-        .root_source_file = .{ .path = "src/tests.zig" },
+        .root_source_file = b.path("src/tests.zig"),
         .target = target,
         .optimize = optimize,
     });
 
-    const run_main_tests = b.addRunArtifact(main_tests);
+    const run_lib_unit_tests = b.addRunArtifact(lib_unit_tests);
 
-    // This creates a build step. It will be visible in the `zig build --help` menu,
+    // Similar to creating the run step earlier, this exposes a `test` step to
-    // and can be selected like this: `zig build test`
+    // the `zig build --help` menu, providing a way for the user to request
-    // This will evaluate the `test` step rather than the default, which is "install".
+    // running the unit tests.
-    const test_step = b.step("test", "Run library tests");
+    const test_step = b.step("test", "Run unit tests");
-    test_step.dependOn(&run_main_tests.step);
+    test_step.dependOn(&run_lib_unit_tests.step);
 }

build.zig.zon

@@ -1,5 +1,72 @@
 .{
-	.name = "hamt",
+    // This is the default name used by packages depending on this one. For
-	.version = "0.1.0",
+    // example, when a user runs `zig fetch --save <url>`, this field is used
-	.dependencies = .{},
+    // as the key in the `dependencies` table. Although the user can choose a
+    // different name, most users will stick with this provided value.
+    //
+    // It is redundant to include "zig" in this name because it is already
+    // within the Zig package namespace.
+    .name = "hamt",
+
+    // This is a [Semantic Version](https://semver.org/).
+    // In a future version of Zig it will be used for package deduplication.
+    .version = "0.1.0",
+
+    // This field is optional.
+    // This is currently advisory only; Zig does not yet do anything
+    // with this value.
+    //.minimum_zig_version = "0.11.0",
+
+    // This field is optional.
+    // Each dependency must either provide a `url` and `hash`, or a `path`.
+    // `zig build --fetch` can be used to fetch all dependencies of a package, recursively.
+    // Once all dependencies are fetched, `zig build` no longer requires
+    // internet connectivity.
+    .dependencies = .{
+        // See `zig fetch --save <url>` for a command-line interface for adding dependencies.
+        //.example = .{
+        //    // When updating this field to a new URL, be sure to delete the corresponding
+        //    // `hash`, otherwise you are communicating that you expect to find the old hash at
+        //    // the new URL.
+        //    .url = "https://example.com/foo.tar.gz",
+        //
+        //    // This is computed from the file contents of the directory of files that is
+        //    // obtained after fetching `url` and applying the inclusion rules given by
+        //    // `paths`.
+        //    //
+        //    // This field is the source of truth; packages do not come from a `url`; they
+        //    // come from a `hash`. `url` is just one of many possible mirrors for how to
+        //    // obtain a package matching this `hash`.
+        //    //
+        //    // Uses the [multihash](https://multiformats.io/multihash/) format.
+        //    .hash = "...",
+        //
+        //    // When this is provided, the package is found in a directory relative to the
+        //    // build root. In this case the package's hash is irrelevant and therefore not
+        //    // computed. This field and `url` are mutually exclusive.
+        //    .path = "foo",
+
+        //    // When this is set to `true`, a package is declared to be lazily
+        //    // fetched. This makes the dependency only get fetched if it is
+        //    // actually used.
+        //    .lazy = false,
+        //},
+    },
+
+    // Specifies the set of files and directories that are included in this package.
+    // Only files and directories listed here are included in the `hash` that
+    // is computed for this package. Only files listed here will remain on disk
+    // when using the zig package manager. As a rule of thumb, one should list
+    // files required for compilation plus any license(s).
+    // Paths are relative to the build root. Use the empty string (`""`) to refer to
+    // the build root itself.
+    // A directory listed here means that all files within, recursively, are included.
+    .paths = .{
+        "build.zig",
+        "build.zig.zon",
+        "src",
+        // For example...
+        //"LICENSE",
+        //"README.md",
+    },
 }

src/bench.zig

@@ -2,10 +2,10 @@ const std = @import("std");
 const HashArrayMappedTrie = @import("hamt").HashArrayMappedTrie;
 
 const StringContext = struct {
-    pub const Digest = u64;
+    pub const Digest = u32;
 
     pub fn hash(input: []const u8) Digest {
-        return std.hash.Wyhash.hash(0, input);
+        return @truncate(std.hash.Wyhash.hash(0, input));
     }
 
     pub fn eql(left: []const u8, right: []const u8) bool {
@@ -17,16 +17,13 @@ const StringArrayHashMap = std.array_hash_map.StringArrayHashMap(void);
 const StringHashMap = std.hash_map.StringHashMap(void);
 
 pub fn main() !void {
-    var gpa = std.heap.GeneralPurposeAllocator(.{}){};
+    const allocator = std.heap.c_allocator;
-    defer std.debug.assert(!gpa.deinit());
-
-    const allocator = gpa.allocator();
     const elem_count = 1000;
 
     const keys = try allocator.alloc([32]u8, elem_count);
     defer allocator.free(keys);
 
-    var rand = std.rand.DefaultPrng.init(1337);
+    var rand = std.rand.DefaultPrng.init(0);
     for (keys) |*key| rand.fill(key);
 
     var trie = try HashArrayMappedTrie([]const u8, void, StringContext).init(allocator);

src/trie.zig

@@ -15,10 +15,10 @@ pub fn HashArrayMappedTrie(comptime K: type, comptime V: type, comptime Context:
 
         const Digest = Context.Digest; // as in Hash Code or Hash Digest
         const table_size = @typeInfo(Digest).Int.bits;
-        const t = @intCast(Log2Int(Digest), @typeInfo(Log2Int(Digest)).Int.bits);
+        const t: Log2Int(Digest) = @intCast(@typeInfo(Log2Int(Digest)).Int.bits);
 
         free_list: FreeList,
-        root: []?*Node,
+        root: *[table_size]?*Node,
 
         const Node = union(enum) { kv: Pair, table: Table };
         const Table = struct { map: Digest = 0, base: [*]Node };
@@ -37,7 +37,7 @@ pub fn HashArrayMappedTrie(comptime K: type, comptime V: type, comptime Context:
                 pub fn deinit(self: *const FreeList.Node, allocator: Allocator, len: usize) void {
                     switch (len) {
                         0 => unreachable,
-                        1 => allocator.destroy(@ptrCast(*Self.Node, self.inner)),
+                        1 => allocator.destroy(@as(*Self.Node, @ptrCast(self.inner))),
                         else => allocator.free(self.inner[0..len]),
                     }
                 }
@@ -45,7 +45,7 @@ pub fn HashArrayMappedTrie(comptime K: type, comptime V: type, comptime Context:
 
             pub fn init(allocator: Allocator) !FreeList {
                 const list = try allocator.create([table_size]?FreeList.Node);
-                std.mem.set(?FreeList.Node, list, null);
+                @memset(list, null);
 
                 return .{ .list = list };
             }
@@ -114,7 +114,7 @@ pub fn HashArrayMappedTrie(comptime K: type, comptime V: type, comptime Context:
             }
 
             pub fn create(self: *FreeList, allocator: Allocator, comptime T: type) !*T {
-                return @ptrCast(*T, try self.alloc(allocator, T, 1));
+                return @ptrCast(try self.alloc(allocator, T, 1));
             }
 
             /// Free'd nodes aren't deallocated, but instead are tracked by a free list where they
@@ -140,14 +140,14 @@ pub fn HashArrayMappedTrie(comptime K: type, comptime V: type, comptime Context:
             }
 
             pub fn destroy(self: *FreeList, allocator: Allocator, node: *Self.Node) !void {
-                self.free(allocator, @ptrCast([*]Self.Node, node)[0..1]);
+                self.free(allocator, @as([*]Self.Node, @ptrCast(node))[0..1]);
             }
         };
 
         pub fn init(allocator: Allocator) !Self {
             // TODO: Add ability to have a larger root node (for quicker lookup times)
-            const root = try allocator.alloc(?*Node, table_size);
+            const root = try allocator.create([table_size]?*Node);
-            std.mem.set(?*Node, root, null);
+            @memset(root, null);
 
             return Self{ .root = root, .free_list = try FreeList.init(allocator) };
         }
@@ -181,10 +181,10 @@ pub fn HashArrayMappedTrie(comptime K: type, comptime V: type, comptime Context:
             }
         }
 
-        fn tableIdx(hash: Digest, offset: u16) Log2Int(Digest) {
+        inline fn tableIdx(hash: Digest, offset: u16) Log2Int(Digest) {
-            const shift_amt = @intCast(Log2Int(Digest), table_size - offset);
+            const shift_amt: Log2Int(Digest) = @intCast(table_size - offset);
 
-            return @truncate(Log2Int(Digest), hash >> shift_amt);
+            return @truncate(hash >> shift_amt);
         }
 
         pub fn search(self: *Self, key: K) ?Pair {
@@ -198,13 +198,12 @@ pub fn HashArrayMappedTrie(comptime K: type, comptime V: type, comptime Context:
                 switch (current.*) {
                     .table => |table| {
                         const mask = @as(Digest, 1) << tableIdx(hash, hash_offset);
+                        if (table.map & mask == 0) return null; // empty table
 
-                        if (table.map & mask != 0) {
+                        const idx = @popCount(table.map & (mask - 1));
-                            const idx = @popCount(table.map & (mask - 1));
+                        current = &table.base[idx];
-                            current = &table.base[idx];
 
-                            hash_offset += t;
+                        hash_offset += t;
-                        } else return null; // hash table entry is empty
                     },
                     .kv => |pair| {
                         if (!Context.eql(pair.key, key)) return null;
@@ -243,7 +242,7 @@ pub fn HashArrayMappedTrie(comptime K: type, comptime V: type, comptime Context:
 
                             var i: Log2Int(Digest) = 0;
                             for (0..table_size) |shift| {
-                                const mask_loop = @as(Digest, 1) << @intCast(Log2Int(Digest), shift);
+                                const mask_loop = @as(Digest, 1) << @as(Log2Int(Digest), @intCast(shift));
 
                                 if (new_map & mask_loop != 0) {
                                     defer i += 1;
@@ -287,7 +286,7 @@ pub fn HashArrayMappedTrie(comptime K: type, comptime V: type, comptime Context:
                                 const copied_pair = try self.free_list.create(allocator, Node);
                                 copied_pair.* = .{ .kv = prev_pair };
 
-                                current.* = .{ .table = .{ .map = mask, .base = @ptrCast([*]Node, copied_pair) } };
+                                current.* = .{ .table = .{ .map = mask, .base = @as([*]Node, @ptrCast(copied_pair)) } };
                             },
                         }
                     },