# fracture 0.2.1

## New features

`fracture()`

and `frac_mat()`

no longer return
an error when receiving a vector containing `NA`

or
`Inf`

values (#14).
`fracture(NA)`

now returns `NA`

.
`fracture(Inf)`

now returns `"Inf/1"`

.

## Bug fixes

- Fixed a bug where
`frac_style()`

would print integers in
mixed `fracture`

s as `"1 ⁰/₀"`

instead of
`"1"`

(#12).

# fracture 0.2.0

## Breaking changes

- The second argument to
`fracture()`

and
`frac_mat()`

is now `...`

, which must be empty. As
a result, all arguments besides `x`

must now be named.
(#5)
`fracture()`

and `frac_mat()`

now default to a
denominator of `1`

when `x`

equals `0`

or `1`

. Previously, these would default to a denominator of
`max_denom`

. `max_denom`

is still used as the
denominator when `x`

is `0`

or `1`

± ε.
(#6)

## New features

`fracture()`

and `frac_mat()`

gain the
argument `denom`

, which allows the user to set an explicit
denominator used by all fractions. (#5)

## Miscellaneous

- The
`print()`

method for `fracture`

s now puts
quotes around mixed fractions to increase legibility. (#7)
- Updated
`testthat`

to 3rd edition. (#5)

# fracture 0.1.3

- Implemented STRICT_R_HEADERS in accordance with
RcppCore/Rcpp#1158

# fracture 0.1.2

## Bug fixes

- Fixed bug where certain fractions
(e.g.
`frac_mat(1307.36, base_10 = TRUE)`

) would cause an
floating point rounding error.
- Reimplemented
`decimal_to_fraction_base_10()`

with
`double`

rather than `int`

.
- Added tests to cover all base_10 fractions below 10,000/10,000
(reduced to 100/100 when testing on CRAN) and a random sample of base_10
fractions below 10,000,000/10,000,000.

## Documentation fixes

`frac_mat()`

’s `max_denom`

should be less than
the *inverse* square root of machine double epsilon.

# fracture 0.1.1

## New features

- Added
`frac_style()`

which formats fractures using
Unicode superscripts and subscripts.
- Added
`is.frac_mat()`

which tests if a matrix is
formatted like the output of `frac_mat()`

.

## Bug fixes

- Fixed bug where certain fractions (e.g. 16/113) would cause an
integer overflow in C++.
- Reimplemented
`decimal_to_fraction_cont()`

with
`double`

rather than `int`

.
- Added tests to cover all fractions below 1000/1000 (reduced to
100/100 when testing on CRAN) and a random sample of fractions below
10,000,000/10,000,000.

- Fixed bug with fracture math where only the first element would be
returned.
- Added error handling for
`NA`

and infinite inputs to
`frac_mat()`

.

## Miscellaneous

# fracture 0.1.0