zig-0-15

name	zig-0.15
description	This skill provides Zig 0.15.x API guidance and should be used when writing or reviewing Zig code. It ensures correct usage of Zig 0.15 APIs, preventing common mistakes from using outdated 0.11/0.12/0.13/0.14 patterns. Essential for ArrayList, std.Io.Writer/Reader (Writergate), HTTP client, Ed25519, JSON, and type introspection APIs.
license	MIT
compatibility	["opencode","claude-code"]
metadata	{"version":"0.15.2","language":"zig","category":"programming-language"}

Zig 0.15.x Programming Guide

Version Scope: This skill is pinned to Zig 0.15.x (specifically 0.15.2). Release Notes: https://ziglang.org/download/0.15.1/release-notes.html

Local Documentation First

Run zig env to discover installation paths — never hardcode them. Key fields: .lib_dir, .std_dir, .version.

• Language Reference: <lib_dir>/../doc/langref.html
• Std Library Source: read files under .std_dir for API verification.

Always check local docs before web search.

Critical API Changes in Zig 0.15

ArrayList (BREAKING)

All mutating methods now require explicit allocator parameter:

// ❌ WRONG (0.13 and earlier)
var list = std.ArrayList(T).init(allocator);
try list.append(item);
try list.appendSlice(items);
_ = try list.addOne();
_ = try list.toOwnedSlice();

// ✅ CORRECT (0.15+)
var list = try std.ArrayList(T).initCapacity(allocator, 16);
defer list.deinit(allocator);
try list.append(allocator, item);
try list.appendSlice(allocator, items);
_ = try list.addOne(allocator);
_ = try list.toOwnedSlice(allocator);

// AssumeCapacity variants do NOT need allocator
list.appendAssumeCapacity(item);

Method	0.13	0.15+
append	try list.append(item)	try list.append(allocator, item)
appendSlice	try list.appendSlice(items)	try list.appendSlice(allocator, items)
addOne	try list.addOne()	try list.addOne(allocator)
ensureTotalCapacity	try list.ensureTotalCapacity(n)	try list.ensureTotalCapacity(allocator, n)
toOwnedSlice	try list.toOwnedSlice()	try list.toOwnedSlice(allocator)
deinit	list.deinit()	list.deinit(allocator)

HashMap

Managed (stores allocator internally):

var map = std.StringHashMap(V).init(allocator);
defer map.deinit();
try map.put(key, value);  // No allocator needed

Unmanaged (requires allocator for each operation):

var map = std.StringHashMapUnmanaged(V){};
defer map.deinit(allocator);
try map.put(allocator, key, value);  // Allocator required

Writergate: New std.Io.Writer and std.Io.Reader (MAJOR REWRITE)

Zig 0.15 replaces the old generic std.io reader/writer with concrete, vtable-based std.Io.Reader / std.Io.Writer. Key differences from the old API:

• Concrete types (not anytype) — can be stored in structs, passed as parameters
• Caller provides buffer — no hidden allocations
• Precise errors — error{ReadFailed} / error{WriteFailed} instead of anyerror
• Ring-buffer design — efficient peek/take without copies

std.Io.Writer

// Core struct (caller provides buffer, vtable handles draining)
pub const Writer = struct {
    vtable: *const VTable,
    buffer: []u8,       // caller-provided; buffer[0..end] = buffered data
    end: usize = 0,
};

Key methods:

Method	Purpose
writeAll(bytes)	Write all bytes (may buffer + drain)
writeByte(byte)	Write single byte
print(fmt, args)	Formatted write
flush()	Drain all buffered data
buffered()	Get slice of currently buffered data

Creating writers:

// 1. Fixed buffer (in-memory, bounded)
var buf: [256]u8 = undefined;
var writer: std.Io.Writer = .fixed(&buf);
try writer.print("count: {d}", .{42});
// written data: buf[0..writer.end]

// 2. Allocating (dynamic, growable)
var aw: std.Io.Writer.Allocating = .init(allocator);
defer aw.deinit();
try aw.writer.print("hello {s}", .{"world"});
const output = aw.written();           // []u8 view
const owned = try aw.toOwnedSlice();   // caller owns

// 3. From file
var file_buf: [4096]u8 = undefined;
var file_writer = file.writer(&file_buf);
try file_writer.interface.writeAll("data\n");
try file_writer.interface.flush();

// 4. stdout / stderr
var stdout_buf: [1024]u8 = undefined;
var stdout_writer = std.fs.File.stdout().writer(&stdout_buf);
const stdout: *std.Io.Writer = &stdout_writer.interface;
try stdout.print("Hello\n", .{});
try stdout.flush();  // Don't forget!

std.Io.Reader

// Core struct (caller provides buffer, vtable handles filling)
pub const Reader = struct {
    vtable: *const VTable,
    buffer: []u8,       // caller-provided; buffer[seek..end] = buffered data
    seek: usize,        // bytes consumed
    end: usize,         // bytes available
};

Key methods:

Method	Purpose
take(n)	Read exactly n bytes, advance position
takeByte()	Read one byte
takeInt(T, endian)	Read integer with endianness
takeDelimiterExclusive(delim)	Read until delimiter (excluded)
takeDelimiterInclusive(delim)	Read until delimiter (included)
peek(n)	Look at next n bytes without consuming
peekByte()	Peek one byte
readSliceAll(buf)	Fill entire buffer from stream
allocRemaining(gpa, limit)	Read all remaining into allocated slice
stream(writer, limit)	Pump data to a Writer
streamRemaining(writer)	Stream all data until EOF
discard(limit)	Skip bytes without reading
buffered()	Get slice of currently buffered data

Creating readers:

// 1. Fixed buffer (in-memory, read-only)
var reader: std.Io.Reader = .fixed("line1\nline2\n");

// 2. From file
var read_buf: [4096]u8 = undefined;
var file_reader = file.reader(&read_buf);
const data = try file_reader.interface.take(10);

// 3. Limited (wrap existing reader with byte limit)
var limit_buf: [256]u8 = undefined;
var limited = reader.limited(.limited(100), &limit_buf);
// limited.interface reads at most 100 bytes

Line-by-line reading:

while (reader.takeDelimiterExclusive('\n')) |line| {
    // process line (delimiter not included)
} else |err| switch (err) {
    error.EndOfStream => {},
    else => return err,
}

Read entire content into allocated slice:

const contents = try reader.allocRemaining(allocator, .limited(1 << 20));
defer allocator.free(contents);

std.Io.Limit

Controls how many bytes a stream/read operation consumes:

const limit = std.Io.Limit.limited(1024);  // at most 1024 bytes
const unlimited = std.Io.Limit.unlimited;   // no limit
const nothing = std.Io.Limit.nothing;       // zero bytes

Reader ↔ Writer: Streaming

// Pump all data from reader to writer
_ = try reader.streamRemaining(&writer);

// Pump with byte limit
_ = try reader.stream(&writer, .limited(4096));

File.Reader / File.Writer

std.fs.File returns wrapper structs; use the .interface field to get the std.Io.Reader / std.Io.Writer:

var write_buf: [4096]u8 = undefined;
var file_writer = file.writer(&write_buf);
try file_writer.interface.writeAll("hello\n");  // .interface is *std.Io.Writer
try file_writer.interface.flush();

var read_buf: [4096]u8 = undefined;
var file_reader = file.reader(&read_buf);
const line = try file_reader.interface.takeDelimiterExclusive('\n');

Two modes: file.writer(buf) (positional, threadsafe) vs file.writerStreaming(buf) (sequential). Prefer positional when available.

Migration from Old std.io

Old (0.14)	New (0.15)
std.io.getStdOut().writer()	std.fs.File.stdout().writer(&buf) → .interface
std.io.fixedBufferStream(&buf)	std.Io.Writer.fixed(&buf) / std.Io.Reader.fixed(data)
fbs.getWritten()	writer.buffered()
writer.writeAll(data)	same, but on *std.Io.Writer
GenericWriter / AnyWriter	*std.Io.Writer (concrete)
GenericReader / AnyReader	*std.Io.Reader (concrete)
reader.readAll(buf)	reader.readSliceAll(buf)
reader.readUntilDelimiter(delim, buf)	reader.takeDelimiterExclusive(delim)

Bridge adapter — if you still have old-style GenericWriter and need new interface:

fn useOldWriter(old_writer: anytype) !void {
    var adapter = old_writer.adaptToNewApi(&.{});
    const w: *std.Io.Writer = &adapter.new_interface;
    try w.print("{s}", .{"example"});
}

HTTP Client (MAJOR REWRITE)

// High-level fetch()
var client: std.http.Client = .{ .allocator = allocator };
defer client.deinit();
const result = try client.fetch(.{
    .location = .{ .url = "https://example.com/api" },
    .method = .GET,
});

// Low-level request/response
const uri = try std.Uri.parse("https://example.com/api");
var req = try client.request(.POST, uri, .{
    .extra_headers = &.{
        .{ .name = "Content-Type", .value = "application/json" },
    },
});
defer req.deinit();
try req.sendBodyComplete(@constCast("{\"key\": \"value\"}"));

var redirect_buf: [4096]u8 = undefined;
var response = try req.receiveHead(&redirect_buf);
// Read body via std.Io.Reader
var body_reader = response.reader(&redirect_buf);
const body = try body_reader.allocRemaining(allocator, .limited(1 << 20));
defer allocator.free(body);

Base64

// ❌ WRONG (0.13) — std.base64.standard.encode / .decode
// ✅ CORRECT (0.15+) — use .Encoder / .Decoder sub-namespace

// Encode
const encoded = std.base64.standard.Encoder.encode(&buf, data);

// Decode
const decoded = try std.base64.standard.Decoder.decode(&out_buf, encoded);

// URL-safe variant: std.base64.url_safe.Encoder / .Decoder

Ed25519 Cryptography (MAJOR CHANGES)

Key types are now structs, not raw byte arrays:

const Ed25519 = std.crypto.sign.Ed25519;

// ❌ WRONG (0.13 - SecretKey was [32]u8)
const secret: [32]u8 = ...;
const kp = Ed25519.KeyPair.fromSecretKey(secret);

// ✅ CORRECT (0.15+ - SecretKey is struct with 64 bytes)
// From 32-byte seed (deterministic):
const seed: [32]u8 = ...;
const kp = try Ed25519.KeyPair.generateDeterministic(seed);

// From 64-byte secret key:
var secret_bytes: [64]u8 = ...;
const secret_key = try Ed25519.SecretKey.fromBytes(secret_bytes);
const kp = try Ed25519.KeyPair.fromSecretKey(secret_key);

// Get public key bytes:
const pubkey_bytes: [32]u8 = kp.public_key.toBytes();

// Get seed:
const seed: [32]u8 = kp.secret_key.seed();

Signature changes:

// ❌ WRONG (0.13 - fromBytes returned error union)
const sig = try Ed25519.Signature.fromBytes(bytes);

// ✅ CORRECT (0.15+ - fromBytes does NOT return error)
const sig = Ed25519.Signature.fromBytes(bytes);

Type Introspection Enums (Case Change)

// ❌ WRONG (0.13 - PascalCase)
if (@typeInfo(T) == .Slice) { ... }
if (@typeInfo(T) == .Pointer) { ... }
if (@typeInfo(T) == .Struct) { ... }

// ✅ CORRECT (0.15+ - lowercase/escaped)
if (@typeInfo(T) == .slice) { ... }
if (@typeInfo(T) == .pointer) { ... }
if (@typeInfo(T) == .@"struct") { ... }
if (@typeInfo(T) == .@"enum") { ... }
if (@typeInfo(T) == .@"union") { ... }
if (@typeInfo(T) == .array) { ... }
if (@typeInfo(T) == .optional) { ... }

Custom Format Functions

// ❌ WRONG (0.13 - used {} format specifier)
pub fn format(
    self: Self,
    comptime fmt: []const u8,
    options: std.fmt.FormatOptions,
    writer: anytype,
) !void {
    _ = fmt;
    _ = options;
    try writer.writeAll("...");
}
// Usage: std.fmt.bufPrint(&buf, "{}", .{value});

// ✅ CORRECT (0.15+ - use {f} format specifier)
pub fn format(self: Self, writer: anytype) !void {
    _ = self;
    try writer.writeAll("...");
}
// Usage: std.fmt.bufPrint(&buf, "{f}", .{value});

JSON Parsing and Serialization

const MyStruct = struct {
    name: []const u8,
    value: u32,
};

// ✅ CORRECT: Parsing (0.15+)
const json_str =
    \\{"name": "test", "value": 42}
;
const parsed = try std.json.parseFromSlice(MyStruct, allocator, json_str, .{});
defer parsed.deinit();
const data = parsed.value;

// ✅ CORRECT: Serialization (0.15.2 - writer-based API)
// Note: There is NO stringifyAlloc in 0.15.2!

// Method 1: Using std.json.Stringify with Allocating writer
var out: std.Io.Writer.Allocating = .init(allocator);
defer out.deinit();
var stringify: std.json.Stringify = .{
    .writer = &out.writer,
    .options = .{},
};
try stringify.write(data);
const json_output = out.written();

// Method 2: Using std.fmt with json.fmt wrapper
const formatted = try std.fmt.allocPrint(allocator, "{f}", .{std.json.fmt(data, .{})});
defer allocator.free(formatted);

// Method 3: To fixed buffer
var buf: [1024]u8 = undefined;
var fbs = std.io.fixedBufferStream(&buf);
var writer = fbs.writer();
var stringify2: std.json.Stringify = .{
    .writer = &writer.adaptToNewApi(&.{}).new_interface,
    .options = .{},
};
try stringify2.write(data);
const output = fbs.getWritten();

// Parse options
const parsed2 = try std.json.parseFromSlice(MyStruct, allocator, json_str, .{
    .allocate = .alloc_always,
    .ignore_unknown_fields = true,
    .max_value_len = 1 << 20,
});

Common Error Messages and Fixes

Error	Cause	Fix
expected 2 argument(s), found 1	ArrayList method missing allocator	Add allocator as first argument
no field or member function named 'encode'	Using std.base64.standard.encode	Use std.base64.standard.Encoder.encode
no field or member function named 'open'	Using old HTTP API	Use client.request() or client.fetch()
expected type 'SecretKey', found '[32]u8'	Ed25519 SecretKey is now struct	Use generateDeterministic(seed)
expected error union type, found 'Signature'	Signature.fromBytes doesn't return error	Remove try
enum has no member named 'Slice'	@typeInfo enum case changed	Use lowercase .slice
no field named 'root_source_file'	Old build.zig API	Use root_module = b.createModule(...)

Build System (build.zig) Changes (MAJOR REWRITE)

Official Guide: https://ziglang.org/learn/build-system/

Executable Creation

// ❌ WRONG (0.14 and earlier)
const exe = b.addExecutable(.{
    .name = "hello",
    .root_source_file = .{ .path = "src/main.zig" },
    .target = target,
    .optimize = optimize,
});

// ✅ CORRECT (0.15+)
const exe = b.addExecutable(.{
    .name = "hello",
    .root_module = b.createModule(.{
        .root_source_file = b.path("src/main.zig"),
        .target = target,
        .optimize = optimize,
    }),
});

Library Creation

// ❌ WRONG (0.14 - addSharedLibrary/addStaticLibrary)
const lib = b.addSharedLibrary(.{
    .name = "mylib",
    .root_source_file = .{ .path = "src/lib.zig" },
});

// ✅ CORRECT (0.15+ - unified addLibrary with linkage)
const lib = b.addLibrary(.{
    .name = "mylib",
    .linkage = .dynamic,  // or .static
    .root_module = b.createModule(.{
        .root_source_file = b.path("src/lib.zig"),
        .target = target,
        .optimize = optimize,
    }),
});

Path Handling

// ❌ WRONG (0.14 - .path field)
.root_source_file = .{ .path = "src/main.zig" },

// ✅ CORRECT (0.15+ - use b.path())
.root_source_file = b.path("src/main.zig"),

Module Dependencies

// ❌ WRONG (0.14)
exe.addModule("sdk", sdk_module);

// ✅ CORRECT (0.15+)
exe.root_module.addImport("sdk", sdk_module);

Testing

// ✅ CORRECT (0.15+)
const unit_tests = b.addTest(.{
    .root_module = b.createModule(.{
        .root_source_file = b.path("src/main.zig"),
        .target = b.graph.host,  // use b.graph.host for native target
    }),
});

const run_unit_tests = b.addRunArtifact(unit_tests);
const test_step = b.step("test", "Run unit tests");
test_step.dependOn(&run_unit_tests.step);

Complete build.zig Example

const std = @import("std");

pub fn build(b: *std.Build) void {
    const target = b.standardTargetOptions(.{});
    const optimize = b.standardOptimizeOption(.{});

    const exe = b.addExecutable(.{
        .name = "hello",
        .root_module = b.createModule(.{
            .root_source_file = b.path("src/main.zig"),
            .target = target,
            .optimize = optimize,
        }),
    });

    b.installArtifact(exe);

    const run_exe = b.addRunArtifact(exe);
    const run_step = b.step("run", "Run the application");
    run_step.dependOn(&run_exe.step);
}

Language Changes

usingnamespace Removed

The usingnamespace keyword has been completely removed in Zig 0.15.

// ❌ WRONG (removed in 0.15)
pub usingnamespace @import("other.zig");

// ✅ CORRECT - explicit re-exports
pub const foo = @import("other.zig").foo;
pub const bar = @import("other.zig").bar;

// ✅ CORRECT - namespace via field
pub const other = @import("other.zig");
// Usage: other.foo, other.bar

Migration for mixins: Use zero-bit fields with @fieldParentPtr:

// ❌ OLD mixin pattern
pub const Foo = struct {
    count: u32 = 0,
    pub usingnamespace CounterMixin(Foo);
};

// ✅ NEW mixin pattern (0.15+)
pub fn CounterMixin(comptime T: type) type {
    return struct {
        pub fn increment(m: *@This()) void {
            const x: *T = @alignCast(@fieldParentPtr("counter", m));
            x.count += 1;
        }
    };
}

pub const Foo = struct {
    count: u32 = 0,
    counter: CounterMixin(Foo) = .{},  // zero-bit field
};
// Usage: foo.counter.increment()

async/await Keywords Removed

The async, await, and @frameSize have been removed. Async functionality will be provided via the standard library's new I/O interface.

// ❌ REMOVED - no async/await keywords
async fn fetchData() ![]u8 { ... }
const result = await fetchData();

// ✅ Use std.Io interfaces or threads instead

Arithmetic on undefined

Operations on undefined that could trigger illegal behavior now cause compile errors:

const a: u32 = 0;
const b: u32 = undefined;

// ❌ COMPILE ERROR in 0.15+
_ = a + b;  // error: use of undefined value here causes illegal behavior

Lossy Integer to Float Coercion

Compile error if integer cannot be precisely represented:

// ❌ COMPILE ERROR in 0.15+
const val: f32 = 123_456_789;  // error: cannot represent precisely

// ✅ CORRECT - opt-in to floating-point rounding
const val: f32 = 123_456_789.0;