Crate awint_internals

This crate contains common developer utilities for crates within the awint system, such as macros that needed a separate crate because #[macro_export] unconditionally causes macros to be publicly accessible. In rare circumstances, someone might want to use the items here for new storage types or highly optimized routines, but most users should never have to interact with this. Be aware that safety requirements can change over time, check bits.rs under awint_core.

There is a hidden reexport of this crate for awint_core, awint_ext, and awint.

Modules

• awi
  Subset of awint::awi

Macros

• const_for
  A basic for loop for const contexts
• forward_debug_fmt
• location
  Returns the Location at the first layer of macros this is in
• subdigits_mut
  Runs f on a digitwise subslice subbits of bits. This is a macro because closures are not properly supported in const functions yet.
• unsafe_binop_for_each
  If lhs.bw() != rhs.bw(), this returns None, otherwise f(x, y) is run on every corresponding pair of digits from first to last.
• unsafe_binop_for_each_mut
  If lhs.bw() != rhs.bw(), this returns None, otherwise f(x, y) is run on every corresponding pair of digits from first to last.
• unsafe_for_each
  f(x) is run on every digit from first to last.
• unsafe_for_each_mut
  f(x) is run on every digit from first to last.

Structs

• CustomDst
  A hack that is very close to the ideal of having a (NonNull<T>, NonZeroUsize) custom DST for Bits to use.
• Location
  Location for an item in the source code. This is essentially a std::panic::Location<'static> with all public fields.
• RawBits
• RawStackBits

Enums

• SerdeError
  A serialization or deserialization error

Constants

• BITS
  Bitwidth of a Digit
• DIGIT_BYTES
  Number of bytes in a Digit
• INV_LB_I1F15
  Reciprocal binary logarithms of the numbers 2..=36 rounded up and in u16p15 fixed point format
• LB_I3F13
  Binary logarithms of the integers 2..=36 rounded up and in u16p13 fixed point format
• MAX
  Maximum value of a Digit
• USIZE_BITS
  Number of bits in a usize

Functions

• bits_upper_bound
  This is used for quickly calculating the maximum number of bits needed for a string representation of a number in some radix to be represented. This may give more bits than needed, but is guaranteed to never underestimate the number of bits needed. Returns None if we see memory exhaustion
• bw
  Utility free function for converting a usize to a NonZeroUsize. This is mainly intended for usage with literals, and shouldn’t be used for fallible conversions.
• chars_upper_bound
  This takes an input of significant bits and gives an upper bound for the number of characters in the given radix needed to represent those bits.
• dd_division
  Divides duo by div and returns the quotient and remainder.
• digits
  Returns the number of whole digits (not including a digit with unused bits) given w
• digits_u
  Returns the number of whole digits (not including a digit with unused bits) given w
• extra
  Returns the number of extra bits given w
• extra_u
  Returns the number of extra bits given w
• panicking_bits_upper_bound
  The same as bits_upper_bound except it panics internally in case of overflow
• panicking_chars_upper_bound
  The same as chars_upper_bound except it panics internally in case of overflow
• total_digits
  Returns the number of Digits needed to represent w, including any digit with unused bits
• widen_add
  Computes x + y + z and returns the widened result as a tuple.
• widen_mul_add
  Computes (x * y) + z. This cannot overflow, because it returns the value widened into a tuple, where the first element is the least significant part of the integer and the second is the most significant.
• widening_mul_add_u128

Type Aliases

• Digit
  The basic element of the internal slice in Bits. This should be a type alias of the unsigned integer of the architecture’s registers. On most architectures, this is simply usize, however there are cases such as AVR where the pointer size is 16 bits but the register size is 8 bits. If this were not register size, it can incur excessive unrolling or underutilization for every loop in the internals.
• IDigit
  Signed version of Digit