power and fix methods

[suleman-uzair]

This PR adds the support power and fix methods for consistency with BigDecimal-Ruby.

related to => plurimath/plurimath-js#28

[ronaldtse]

@hmdne are you okay with this? Thanks!

[hmdne]

Looks good to me. I will backport it to the opal/opal repo and try to see if we can enable some more rubyspec tests.

[hmdne]

In short, all is fine, a couple of additional Rubyspec tests are passing now (https://github.com/ruby/spec/tree/ee83ee683bf15bb2bdd98498069f298abefbdbcc is the repo):

  fails "BigDecimal#** 0 to power of 0 is 1" # NoMethodError: undefined method `**' for 0
  fails "BigDecimal#** 0 to powers < 0 is Infinity" # NoMethodError: undefined method `**' for 0
  fails "BigDecimal#** other powers of 0 are 0" # NoMethodError: undefined method `**' for 0
  fails "BigDecimal#** powers of 1 equal 1" # NoMethodError: undefined method `**' for 1
  fails "BigDecimal#** returns 0.0 if self is infinite and argument is negative" # NoMethodError: undefined method `**' for Infinity
  fails "BigDecimal#** returns NaN if self is NaN" # NoMethodError: undefined method `**' for NaN
  fails "BigDecimal#** returns infinite if self is infinite and argument is positive" # NoMethodError: undefined method `**' for Infinity
  fails "BigDecimal#fix correctly handles special values" # NoMethodError: undefined method `fix' for Infinity
  fails "BigDecimal#fix does not allow any arguments" # Expected ArgumentError but got: NoMethodError (undefined method `fix' for 1.23456789)
  fails "BigDecimal#fix returns 0 if the absolute value is < 1" # NoMethodError: undefined method `fix' for 0.99999
  fails "BigDecimal#fix returns a BigDecimal" # NoMethodError: undefined method `fix' for Infinity
  fails "BigDecimal#fix returns the integer part of the absolute value" # NoMethodError: undefined method `fix' for 2e+1000
  fails "BigDecimal#power 0 to power of 0 is 1" # NoMethodError: undefined method `power' for 0
  fails "BigDecimal#power 0 to powers < 0 is Infinity" # NoMethodError: undefined method `power' for 0
  fails "BigDecimal#power other powers of 0 are 0" # NoMethodError: undefined method `power' for 0
  fails "BigDecimal#power powers of 1 equal 1" # NoMethodError: undefined method `power' for 1
  fails "BigDecimal#power returns 0.0 if self is infinite and argument is negative" # NoMethodError: undefined method `power' for Infinity
  fails "BigDecimal#power returns NaN if self is NaN" # NoMethodError: undefined method `power' for NaN
  fails "BigDecimal#power returns infinite if self is infinite and argument is positive" # NoMethodError: undefined method `power' for Infinity

But also there are new failures, which probably were disabled due to missing functions, some of which you may want to correct for better compatibility.

fails "BigDecimal#truncate returns NaN if self is NaN" # ArgumentError: [BigDecimal#fix] wrong number of arguments (given 1, expected 0)
fails "BigDecimal#truncate returns value of type Integer." # Expected false == true to be truthy but was false
fails "BigDecimal#truncate sets n digits left of the decimal point to 0, if given n < 0" # ArgumentError: [BigDecimal#fix] wrong number of arguments (given 1, expected 0)

[suleman-uzair]

@hmdne, I have fixed the argument error, and now there is an error related to BigNumber’s support for negative integers as the truncate method’s argument.
As far as I could explore the BigNumber v2.1.4, the minimum integer allowed for round (called in truncate) is 0, while BigDecimal supports negative integers.

for example In Ruby:

BigDecimal("1212.11").truncate(-1).to_s("F") # "1210.0"

But in BigNumber this will raise the following error (as mentioned here):

decimal places out of range

can we skip the negative integers support for now or do we need this implementation?

cc: @ronaldtse