Bug #13848: BigDecimal.new('200.') raises an exception

• Status: Closed
• Assignee: mrkn (Kenta Murata)
• Target version: -
• ruby -v: ruby 2.4.1p111 (2017-03-22 revision 58053) [x86_64-linux]
• Backport: 2.2: UNKNOWN, 2.3: UNKNOWN, 2.4: UNKNOWN

[ddoherty03 (Daniel Doherty)]

BigDecimal.new('200.') raises an exception:

ArgumentError: invalid value for BigDecimal(): "200."

It should accept this as a valid decimal number.

This requires fixing numbers read in from the wild.

[shevegen (Robert A. Heiler)]

I believe you mean for BigDecimal.new('200.') to be equivalent to BigDecimal.new('200.0')?

I could see both ways be ok; there may be use cases where the error is better because of
the user not being explicit enough; or the user does not want to have to type in the
trailing '0'. So I can't say whether this is really a bug or not, it may be specifically
wanted to be that way.

I guess if there are no trade offs though, then I'd agree with the proposal if you meant
that it is the same to mean for a trailing '0' there - but I could also see for purity
to require the '0' to be passed.

What about input such as:

BigDecimal.new('0.')

I mean it will be 0 anyway... it reads strangely to omit the the '0' to my eyes though ...

Latest docu can be found here, if anyone else is interested:

https://ruby-doc.org/stdlib/libdoc/bigdecimal/rdoc/BigDecimal.html

[ddoherty03 (Daniel Doherty)]

Robert, yes, I think it should be the same as BigDecimal.new('200.0'). The problem arises, not when a programmer is typing the number---requiring a programmer to be explicit is OK by me---but when strings are processed that come from a user of the code.

ss = '200.' # Read from a user-prepared file.
BigDecimal.new(ss) => ArgumentError ...

As it is, the programmer has to massage every string to recognize a reasonable, albeit a bit sloppy, string that is clearly meant as a valid number.

Also, it seems crazy to me that

BigDecimal.new('200hello world') => 0.2e3

but BigDecimal('200.') raises an ArgumentError exception. Just sayin...

Regards,

[kernigh (George Koehler)]

Float('200.') also raises ArgumentError. In this way, BigDecimal.new seems consistent with Float. But it's strange that Float('200hello world') raises ArgumentError but BigDecimal.new('200hello world') succeeds.

[mrkn (Kenta Murata)]

• Status changed from Open to Assigned
• Assignee set to mrkn (Kenta Murata)
• Target version set to 2.6

I'm working this bug at https://github.com/ruby/bigdecimal/pull/87

[naruse (Yui NARUSE)]

• Target version deleted (2.6)

[mrkn (Kenta Murata)]

• Status changed from Assigned to Closed

Fixed in r66222