Issue #14588 has been updated by nobu (Nobuyoshi Nakada).

Status changed from Open to Rejected

Anon92929 (Anon Ymous) wrote:
> NaN should not be an exception. No form of NaN should an exception, it should just be NaN.

Do you mean "quiet NaN"?

Ruby uses signaling NaN by "fail-earlier" design.
All calculations which NaN involves result in NaN, and it's hard to tell from where it came.
Maybe it could be enough to fallback into NaN in your particular use case, but it isn't in general.


----------------------------------------
Bug #14588: math library functions should NOT raise exceptions
https://bugs.ruby-lang.org/issues/14588#change-71345

* Author: Anon92929 (Anon Ymous)
* Status: Rejected
* Priority: Normal
* Assignee: 
* Target version: 
* ruby -v: 
* Backport: 2.3: UNKNOWN, 2.4: UNKNOWN, 2.5: UNKNOWN
----------------------------------------
BigDecimal, Integer, Float, none of these should raise exceptions, but they should all fail SILENTLY or return NaN during error cases.

I shouldn't have to go fix every possible way that BigDecimal might throw a breaking change in a thousand places in my codebase. I need math libraries that DON'T BREAK!!!


BUG RESUBMITTED. 

The bug was rejected at github because I was referred to open a ticket at ruby-lang.org.

The bug was rejectet at ruby-lang.org because the ticket was closed at github.

Not cool, guys.



-- 
https://bugs.ruby-lang.org/

Unsubscribe: <mailto:ruby-core-request / ruby-lang.org?subject=unsubscribe>
<http://lists.ruby-lang.org/cgi-bin/mailman/options/ruby-core>