v2.0.0

Version 2.0.0 of TypeUtils (2026-06-12)

Breaking changes

• By default, adapt_precision only converts non-integer numbers. The former behavior wasto convert all numbers which is not desirable as a general rule for integers (which include Booleans). The fixed behavior is that an integer or complex number with integer real and imaginary parts is left unchanged by adapt_precision while a rational number is converted to floating-point. It is still possible to overcome this behavior for a custom type by specializing adapt_precision.

Changed

• get_precision can work on any types and instances.

• Minimal Julia version is 1.0, this bound is imposed by the InverseFunctions package.

Added

• Function adapt_multiplier_precision(α, x) to adapt the precision of a scalar numerical multiplier α to that of the other operand x (an array).

• is_unitless(x) returns whether number x (a numeric value or type) is a unitless number.

• is_static_number(x) returns whether number x (a numeric value or type) is a static number, i.e. a number whose value is known considering its type.

v1.14.0

TypeUtils v1.14.0

Diff since v1.13.0

This new minor version essentially provides lazymap (see below).

Added

• B = lazymap([T::Type,], f, A[, f_inv]) yields an object B that lazily maps function f to array or iterator A. Optional argument T is the element-type of B which is automatically inferred if not specified. If A is an (abstract) array, f_inv may be provided to specify the inverse of f. Related type alias LazyMap{T,N,F,typeof(A)} can be used to specialize on the type B with N = ndims(A) and F = typeof(f).

Change

• The former function as_eltype(T,A) is now a shortcut for lazymap(T,identity,A) unless T == eltype(A) holds in which case A is returned.

• Private type TypeUtils.AsEltype (formerly used by as_eltytpe) no longer exists.

v1.13.0

TypeUtils v1.13.0

Diff since v1.12.0

Added

• units_of(x) and units_of(typeof(x)) yield the units of the number x. The units are a trait which only depends on the type of the argument. This function is a replacement for Unitful.unit when this package is not yet loaded.

Merged pull requests:

• Bump actions/checkout from 5 to 6 (#4) (@dependabot[bot])

v1.12.0

TypeUtils v1.12.0

Diff since v1.11.0

Fixed

• get_precision on n-tuples yields a correct result and is inferable.

Added

• get_precision() yields AbtractFloat instead of throwing a MethodError.

v1.11.0

TypeUtils v1.11.0

Diff since v1.10.0

Fixed

• Fix tests for Julia 1.12 and above.

• Syntax @public a, b, c is allowed.

Merged pull requests:

• Bump actions/checkout from 4 to 5 (#3) (@dependabot[bot])

v1.10.0

TypeUtils v1.10.0

Diff since v1.9.0

Added

• With multiple arguments, get_precision(x, y...) yields the best precision of all the given arguments.

• nearest(T, x) can deal with dimensionful numbers and types.

v1.9.0

TypeUtils v1.9.0

Diff since v1.8.0

This new minor version adds methods to deal with the numerical precision of numbers and objects.

Added

• get_precision(x) yields the numerical precision of x.

• adapt_precision(T, x) yields a version of x with numerical precision T.

• destructure(Vector, x) and destructure(Vector{T}, x) yield a vector while destructure(Tuple, x) yields a tuple as destructure(x).

v1.8.0

TypeUtils v1.8.0

Diff since v1.7.1

Added

• convert_eltype(T,A) for A an instance of LinearAgebra.Factorization.

• Tests with Aqua.jl.

Fixed

• A few minor things detected by Aqua.jl have been fixed.

• Better rule for element type conversion of AbstractUnitRange instances.

• Fix specialization of Base.return_types which shall return a vector of types.

v1.7.1

TypeUtils v1.7.1

Diff since v1.7.0

Fixed

• is_signed(x) is false if x is a complex number with unsigned rational parts.

v1.7.0

TypeUtils v1.7.0

Diff since v1.6.0

• New is_signed method to infer whether a value of a given numeric type can be negated while retaining the same type.
• Extend similar for the type of arrays returned by as_eltype.