Bug #9689
closed(BigDecimal + Float) no longer commutative due to #7176
Description
So, there was a bug filed a while back (#7176) about Rational + BigDecimal failing, despite BigDecimal + Rational working.
https://bugs.ruby-lang.org/issues/7176
When this bug was fixed (in r37406), the new BigDecimal#+ also (I believe unnecessarily) converts Float to BigDecimal, making BigDecimal + Float return a BigDecimal despite Float + BigDecimal returning a Float. Given that work was put in to make BigDecimal + Rational commutative it seems like the same behavior should be in place for BigDecimal + Float: preferably, either both should be commutative, or neither should be commutative. (Please understand that I filed this as a "Bug", as this seems like a regression from the earlier behavior, but I would not at all quibble if someone decided that this should actually be classified as a "Feature" instead.)
(BTW, while doing due diligence on this issue, I determined that there was a previous bug #2129 that was filed against Rational + BigDecimal years ago that has almost the same description as 7176. At the time, Rational + BigDecimal returned a Rational; I think this is the more correct behavior, because any BigDecimal can be represented as a Rational, but not all Rationals can be represented as a BigDecimal. As a concrete example, 1/9 has no BigDecimal representation. Given that this behavior has already changed before, it seems like a fix for the commutativity of Float+BigDecimal should probably also deal with this other problem.)
Updated by mrkn (Kenta Murata) almost 10 years ago
- Assignee set to mrkn (Kenta Murata)
Updated by zzak (zzak _) almost 10 years ago
- Category set to ext
- Status changed from Open to Assigned
- Target version set to 2.2.0
Updated by mrkn (Kenta Murata) over 7 years ago
- Status changed from Assigned to Closed
This is fixed in the latest master branch of ruby/bigdecimal.