Fix floating formatting

[marocchino]

close #58

[marocchino]

> BigDecimal(10).to_s('F')
=> "10.0"

However, the original test expects the result to be 10, so I added a conditional branch to handle this.

[marocchino]

$ cat bench.rb
# frozen_string_literal: true

require 'benchmark/ips'
require 'bigdecimal'

def format_sub(num)
  format('%f', num).sub(/\.?0+$/, '')
end

def big_decimal(num)
  BigDecimal(num).to_s('F')
end

Benchmark.ips do |x|
  x.report('format_sub') { format_sub(1.23456789) }
  x.report('big_decimal') { big_decimal(1.23456789) }
  x.compare!
end
$ ruby bench.rb
ruby 3.4.1 (2024-12-25 revision 48d4efcb85) +PRISM [arm64-darwin24]
Warming up --------------------------------------
          format_sub   126.984k i/100ms
         big_decimal   164.232k i/100ms
Calculating -------------------------------------
          format_sub      1.273M (± 0.7%) i/s  (785.70 ns/i) -      6.476M in   5.088579s
         big_decimal      1.638M (± 0.7%) i/s  (610.61 ns/i) -      8.212M in   5.014346s

Comparison:
         big_decimal:  1637706.2 i/s
          format_sub:  1272746.5 i/s - 1.29x  slower

[sethdeckard]

Hi @marocchino thank you for the PR, would you please address the following?

• Fix spec.dependencies (should be add_dependency)
• Move float_format to AttributeFormatter (looks like we'll need it for other types that have this same issue and I thought this was better than making it private).

The same issue exists in PartItem and DateRangeItem via unquoted_format but I can handle this if you just want to keep this PR focused, up to you.

[sethdeckard]

LGTM, thank you!

[sethdeckard]

@marocchino released in 1.8.1