What's easiest way to compare a Float & BigDecimal (i.e. like a equals mechanism)


#1

Hi,

what’s easiest way to compare a Float & BigDecimal (i.e. like a equals
mechanism)? It seems that “==” does NOT work as they are different
types. Is there a way to do a comparison without having to convert
types?

?> a
=> #BigDecimal:2395e98,‘0.65E1’,8(8)

x
=> 6.5

a == x
=> false #<<== DIDN’T WORK

tks


#2

Greg H. wrote:

=> 6.5

a == x
=> false #<<== DIDN’T WORK

tks

It works fine for me. If I do something like this:

f = 0.1
b = BigDecimal.new(‘0.1’)
puts ‘yes’ if f == b

It will return true. I’ve always stayed clear of the equality operator
with floating point numbers. Floating point cannot represent all
numbers exactly, so comparing it with something like a BigDecimal (which
can represent all numbers exactly) might be problematic.


#3

that works - but my 6.5 number doesn’t work - does the following also
not work for you?

a = BigDecimal.new(‘6.5’)
=> #BigDecimal:238875c,‘0.65E1’,8(8)

b = 6.5
=> 6.5

a == b
=> false

a.class
=> BigDecimal

b.class
=> Float


#4

On Sat, Nov 8, 2008 at 2:44 PM, Greg H.
removed_email_address@domain.invalid wrote:

=> BigDecimal

b.class
=> Float

You can rely on comparing floating point numbers for less than or
greater than only.
You can not reliably compare floating point numbers for equality period.


#5

On Nov 8, 8:26 am, Greg H. removed_email_address@domain.invalid
wrote:

Hi,

what’s easiest way to compare a Float & BigDecimal (i.e. like a equals
mechanism)? It seems that “==” does NOT work as they are different
types. Is there a way to do a comparison without having to convert
types?

Actually, there is none. Why don’t you just let Ruby do the job and
multiply the bigdecimal with 1.0 when comparing? Also, as previously
noted, == operator with floats is not … reliable.

br,

pen


#6

oh so convert the BigDecimal to a float by multiplying by 1.0 you mean
first?


#7

On Nov 8, 1:26 am, Greg H. removed_email_address@domain.invalid
wrote:

a == x

=> false #<<== DIDN’T WORK

I thought I saw a similar discussion not too long ago… aha:

http://groups.google.com/group/comp.lang.ruby/msg/bc1e54150d1d45d5

With the approx_equal? method in that thread, this works:

a = BigDecimal.new(‘6.5’)
f = 6.5
f == a
#=> false
f.approx_equal?(a, 0.00000000001)
#=> true

– Mark.


#8

Greg H. wrote:

types?
Actually, there is none. Why don’t you just let Ruby do the job and
multiply the bigdecimal with 1.0 when comparing? Also, as previously
noted, == operator with floats is not … reliable.

br,

pen

BigDecimal objects should have a to_f method. Using that would probably
be more clear than multiplying by 1.0 (a seamingly meaningless thing to
do).