bignum 0.3.0

• New seq.bignum_vctr() for generating sequences of biginteger or bigfloat (#30).

• To suppress lossy cast warnings, you should now use suppressWarnings() (#29).

  • If using R 4.1+, we recommend suppressWarnings(expr, classes = "bignum_warning_cast_lossy").
  • Compatible with rlang 1.0.0.

• vignette("operations") has been removed and this documentation has been moved to man topics.

bignum 0.2.2

• Adapted tests for testthat 3.1.0 (#27).

bignum 0.2.1

• format() no longer checks for misspelled arguments. This previously caused issues when storing a bignum vector inside a data.frame or data.table (#25).

bignum 0.2.0

New features

• New vignette("type-hierarchy") and vignette("precision").
• format() functions now support customized output.
  • New sigfig and digits arguments control the displayed precision.
  • New notation argument chooses decimal, scientific or hexadecimal output.
  • New options "bignum.sigfig" and "bignum.max_dec_width" determine the default formatting.
• When a bignum vector is stored in a tibble column, the formatting is adjusted to aid reading data vertically.
  • The decimal point and exponent are aligned across rows and negative numbers are colored red.
  • The options "pillar.sigfig" and "pillar.max_dec_width" determine tibble formatting.
  • See vignette("digits", package = "pillar") for details.
• digamma() and trigamma() operations are now supported.
• log() now supports the base argument.

Bug fixes

• Casting a non-integer double() to biginteger() now returns the truncated integer, consistent with base vectors. Previously NA was returned. A lossy cast warning is still raised.
• Casting a large double() to biginteger() now works correctly. Previously NA was returned, depending on the value of options("scipen").
• Casting Inf to biginteger() now raises a lossy cast warning.
• Casting a large biginteger() to bigfloat() now raises a lossy cast warning when the bigfloat() precision is exceeded.
• is.finite() and is.infinite() now correctly handle large bigfloat() values. Previously large values were considered infinite.
• Comparisons between biginteger() and double() vectors are now possible (e.g. biginteger(2) > 1.5). Previously a lossy cast error was raised.

bignum 0.1.0

First CRAN release.

• Numeric vector classes:

  • biginteger() stores any integer (i.e. arbitrary precision).
  • bigfloat() stores 50 decimal digits of precision.

• Constants for common situations: NA_biginteger_, NA_bigfloat_, bigpi.

• Support for many basic operations (see vignette("operations")):

  • Check for special values: is.na(), is.finite(), is.infinite(), is.nan()
  • Comparison: <, >, <=, >=, ==, !=
  • Arithmetic: +, -, *, /, ^, %%, %/%
  • Mathematical:
    • sum(), prod(), max(), min(), range(), mean()
    • cumsum(), cumprod(), cummax(), cummin()
    • floor(), ceiling(), trunc()
    • abs(), sign(), sqrt()
    • log(), log10(), log2(), log1p(), exp(), expm1()
    • cos(), sin(), tan(), acos(), asin(), atan(), cospi(), sinpi(), tanpi()
    • cosh(), sinh(), tanh(), acosh(), asinh(), atanh()
    • gamma(), lgamma(), factorial(), lfactorial()