# zig_units **Compile-time dimensional analysis for Zig.** `zig_units` lets you attach physical units to numeric values so that dimension mismatches become *compile errors* rather than silent bugs. At runtime a `Quantity` is nothing but a single number — zero overhead. ``` velocity = distance / time → L¹T⁻¹ ✓ force = mass + velocity → compile error: M¹ ≠ L¹T⁻¹ ``` Requires **Zig 0.16** or later. --- ## Features - Seven SI base dimensions (`L M T I Tp N J`) - Full SI prefix support (`P T G M k h da d c m u n p f`) - Custom time aliases (`.min`, `.hour`, `.year`) - Automatic scale conversion on add/sub (finer unit wins) - `Quantity(T, dims, scales)` — scalar, any numeric backing type - `QuantityVec3` — three-component vector with the same guarantees - Unicode superscript formatting (`9.81m.s⁻²`) - Integer-safe square root for `Vec3.length()` - All dimension tracking happens at `comptime` — no runtime cost --- ## Installation ### Add as a Zig dependency ```bash zig fetch --save https://github.com/YOUR_USERNAME/zig_units/archive/refs/heads/main.tar.gz ``` This adds an entry to your `build.zig.zon`. Then wire it up in your `build.zig`: ```zig const zig_units = b.dependency("zig_units", .{ .target = target, .optimize = optimize, }); my_module.addImport("zig_units", zig_units.module("zig_units")); ``` ### Local path (monorepo / development) ```zig // build.zig.zon .dependencies = .{ .zig_units = .{ .path = "../zig_units" }, }, ``` --- ## Quick start ```zig const units = @import("zig_units"); const Quantity = units.Quantity; const Dims = units.Dimensions; const Scales = units.Scales; // Define named unit types const Meter = Quantity(f32, Dims.init(.{ .L = 1 }), Scales.init(.{})); const KiloMeter= Quantity(f32, Dims.init(.{ .L = 1 }), Scales.init(.{ .L = .k })); const Second = Quantity(f32, Dims.init(.{ .T = 1 }), Scales.init(.{})); const MPerSec = Quantity(f32, Dims.init(.{ .L = 1, .T = -1 }), Scales.init(.{})); const dist = Meter{ .value = 100.0 }; const t = Second{ .value = 5.0 }; // Dimension is tracked automatically — vel has type L¹T⁻¹ const vel = dist.divBy(t); // Convert to an explicit type (same dims required, compile error otherwise) const vel2 = vel.to(MPerSec); // Cross-scale addition: km + m → result in metres (finer scale) const km = KiloMeter{ .value = 1.0 }; const sum = km.add(dist); // 1100 m ``` --- ## API reference ### `Quantity(T, dims, scales)` | Member | Kind | Description | |---|---|---| | `value` | field | The raw numeric value | | `ValueType` | comptime | Alias for `T` | | `dims` | comptime | The `Dimensions` of this type | | `scales` | comptime | The `Scales` of this type | | `Vec3` | comptime | The matching `QuantityVec3` type | | `add(rhs)` | fn | Same-dimension addition, finer scale | | `sub(rhs)` | fn | Same-dimension subtraction, finer scale | | `mulBy(rhs)` | fn | Multiplication, dims are added | | `divBy(rhs)` | fn | Division, dims are subtracted | | `scale(s: T)` | fn | Dimensionless scalar multiply | | `to(Dest)` | fn | Convert to another `Quantity` type (same dims) | | `vec3()` | fn | Broadcast scalar to a `Vec3` | | `format(writer)` | fn | Print `value + unit string` | ### `QuantityVec3` Obtained via `SomeQuantity.Vec3`. | Member | Kind | Description | |---|---|---| | `x, y, z` | fields | The three components | | `zero` | comptime | `(0, 0, 0)` | | `one` | comptime | `(1, 1, 1)` | | `initDefault(v)` | fn | Broadcast scalar to all components | | `add(rhs)` | fn | Component-wise addition | | `sub(rhs)` | fn | Component-wise subtraction | | `mulBy(rhs)` | fn | Component-wise element-wise multiply | | `divBy(rhs)` | fn | Component-wise element-wise divide | | `mulByScalar(q)` | fn | Multiply by a scalar `Quantity` | | `divByScalar(q)` | fn | Divide by a scalar `Quantity` | | `scale(s: T)` | fn | Dimensionless scalar multiply | | `negate()` | fn | Negate all components | | `to(DestQ)` | fn | Convert to another vector quantity type | | `lengthSqr()` | fn | Squared Euclidean length (no sqrt) | | `length()` | fn | Euclidean length (integer-safe) | | `format(writer)` | fn | Print `(x, y, z) + unit string` | ### `Dimensions` A comptime struct storing a signed exponent per SI base dimension. ```zig const Dims = @import("zig_units").Dimensions; // Acceleration: L¹ T⁻² const accel_dims = Dims.init(.{ .L = 1, .T = -2 }); ``` | Function | Description | |---|---| | `init(struct_literal)` | Create from named exponents; unset dims default to 0 | | `initFill(val: i8)` | Set all exponents to `val` | | `get(dim)` | Read a single exponent | | `set(dim, val)` | Write a single exponent | | `add(a, b)` | Component-wise sum (for `mulBy`) | | `sub(a, b)` | Component-wise difference (for `divBy`) | | `eql(a, b)` | Equality check | | `str()` | Comptime human-readable string, e.g. `"L1T-2"` | ### `Scales` A comptime struct storing a `UnitScale` per SI base dimension. ```zig const Scales = @import("zig_units").Scales; // Kilometres per nanosecond const spd_scales = Scales.init(.{ .L = .k, .T = .n }); ``` | `UnitScale` variant | Factor | |---|---| | `.P` | ×10¹⁵ | | `.T` | ×10¹² | | `.G` | ×10⁹ | | `.M` | ×10⁶ | | `.k` | ×10³ | | `.h` | ×10² | | `.da` | ×10¹ | | `.none` | ×1 | | `.d` | ×10⁻¹ | | `.c` | ×10⁻² | | `.m` | ×10⁻³ | | `.u` | ×10⁻⁶ | | `.n` | ×10⁻⁹ | | `.p` | ×10⁻¹² | | `.f` | ×10⁻¹⁵ | | `.min` | ×60 (seconds) | | `.hour` | ×3 600 | | `.year` | ×31 536 000 | --- ## Examples ### Kinematics ```zig const Meter = Quantity(f64, Dims.init(.{ .L = 1 }), Scales.init(.{})); const Second = Quantity(f64, Dims.init(.{ .T = 1 }), Scales.init(.{})); const pos = Meter{ .value = 200.0 }; const time = Second{ .value = 8.0 }; const vel = pos.divBy(time); // L¹T⁻¹ — 25 m/s const accel = vel.divBy(time); // L¹T⁻² — 3.125 m/s² ``` ### Cross-scale addition ```zig const KM = Quantity(i64, Dims.init(.{ .L = 1 }), Scales.init(.{ .L = .k })); const M = Quantity(i64, Dims.init(.{ .L = 1 }), Scales.init(.{})); const a = KM{ .value = 2 }; // 2 km const b = M{ .value = 500 }; // 500 m const sum = a.add(b); // result scale = metres (finer) → 2500 m ``` ### Time conversion ```zig const Hour = Quantity(i64, Dims.init(.{ .T = 1 }), Scales.init(.{ .T = .hour })); const Minute = Quantity(i64, Dims.init(.{ .T = 1 }), Scales.init(.{ .T = .min })); const Second = Quantity(i64, Dims.init(.{ .T = 1 }), Scales.init(.{})); const h = Hour{ .value = 2 }; const min = h.to(Minute); // 120 const sec = min.to(Second); // 7200 ``` ### Vec3 velocity ```zig const Meter = Quantity(f32, Dims.init(.{ .L = 1 }), Scales.init(.{})); const Second = Quantity(f32, Dims.init(.{ .T = 1 }), Scales.init(.{})); const pos = Meter.Vec3{ .x = 30.0, .y = 60.0, .z = 90.0 }; const time = Second{ .value = 3.0 }; const vel = pos.divByScalar(time); // Vec3 with dims L¹T⁻¹ const dist = vel.length(); // Euclidean length ``` ### Dimension mismatch — compile error ```zig const Meter = Quantity(f32, Dims.init(.{ .L = 1 }), Scales.init(.{})); const Second = Quantity(f32, Dims.init(.{ .T = 1 }), Scales.init(.{})); const d = Meter{ .value = 5.0 }; const t = Second{ .value = 2.0 }; // This will NOT compile: const bad = d.add(t); // error: Dimension mismatch in add: L1 vs T1 ``` --- ## Running the tests ```bash zig build test ``` The test suite covers scalar and vector arithmetic, cross-scale operations, conversion chains, negative values, formatting, and an optional benchmark (`"Comprehensive Benchmark: All Ops × All Types"`). --- ## Project layout ``` zig_units/ ├── build.zig # Build script; exposes the "zig_units" module ├── build.zig.zon # Package manifest ├── src/ │ ├── main.zig # Quantity, QuantityVec3, tests │ ├── Dimensions.zig # SI base dimensions + comptime arithmetic │ ├── Scales.zig # SI prefixes + scale helpers │ └── helper.zig # Internal utilities (isInt, printSuperscript) └── README.md ``` --- ## Design notes **Why comptime parameters?** Zig's `comptime` means the compiler can evaluate all dimension arithmetic before any machine code is generated. Two quantities with mismatched dimensions simply fail to compile — there is no runtime overhead and no need for exception handling. **Scale selection on arithmetic.** When two operands have different scales (e.g. km and m), `zig_units` automatically picks the finer (smaller-factor) scale for the result. This prevents silent precision loss at the cost of an automatic rescaling of both operands. **Integer backing types.** Division uses an `f64` intermediate and rounds back to the integer type. For best accuracy, prefer `f32`/`f64` for quantities that will be divided frequently. --- ## License MIT — see `LICENSE` for details.