AW:BigDecimal - bug?

ruby-talk
1

Conversion can be done with BigDecimal.new

A Berger <aberger7890@gmail.com> schrieb:

···

Unsubscribe: <mailto:ruby-talk-request@ruby-lang.org?subject=unsubscribe>
<http://lists.ruby-lang.org/cgi-bin/mailman/options/ruby-talk&gt;

2

Hi
although this doesnt work:
r=4/3.to_r
f=r.to_f

p BigDecimal.new(r)
p BigDecimal.new(f)
# 2x error!
no direct conversion possible.

Berg

3

This worked for me:

    r = Rational(4, 3)
    # or in ruby 2.1+:
    #r = 4/3r

    f = r.to_f

    require 'bigdecimal'
    BigDecimal.new r, 20
    #=> #<BigDecimal:2223a88,'0.1333333333 3333333333E1',36(45)>

    BigDecimal.new f, 10
    # => #<BigDecimal:22205e0,'0.1333333333E1',18(36)>

Omitting the second parameter resulted in errors like: "can't omit
precision for a Rational." but that was a no-brainer to resolve in
combination with:

···

On 28/02/2016 7:25 PM, "A Berger" <aberger7890@gmail.com> wrote:

Hi
although this doesnt work:
r=4/3.to_r
f=r.to_f

p BigDecimal.new(r)
p BigDecimal.new(f)
# 2x error!
no direct conversion possible.

Berg

4

Hi

Omitting the second parameter resulted in errors like: "can't omit

precision for a Rational." but that was a no-brainer to resolve in
combination with:

I have read that docu (before), but testet only the case without digits:
" ... If omitted or 0, the number of significant digits is determined from
the initial value."
Why does it not work with float? (Ok, I didnt understand the error-msg,
because first tested it with float(no rational involved!)

So its a bug - in most cases I dont want to specify digits.
But if it works with digits, it is manageable (shouldnt be hard to correct
the module: digits=x.to_f.length-1 (for the dot) ? )

Thanks for this hint. I gave up too early.
Berg