ruby-core

Issue #9575 has been updated by Marc-Andre Lafortune.


Found many other problems when some arguments are floats:

    4.2.step(by: 0.0).first(2)           # => [], should be [4.2, 4.2]
    4.2.step(by: -0.0).first(2)          # => [-Infinity, -Infinity], should be [4.2, 4.2]
    42.step(by: 0, to: -Float::INFINITY).first(2) # => [], should be [42, 42]
    42.step(by: 0, to: 42.5).first(2)    # => [42.0, 42.0], should really be [42, 42]

I believe I've fixed all cases.

----------------------------------------
Bug #9575: Step with 0 step is buggy
https://bugs.ruby-lang.org/issues/9575#change-45509

* Author: Marc-Andre Lafortune
* Status: Open
* Priority: Normal
* Assignee: Marc-Andre Lafortune
* Category: core
* Target version: current: 2.2.0
* ruby -v: r45199
* Backport: 1.9.3: UNKNOWN, 2.0.0: UNKNOWN, 2.1: UNKNOWN
----------------------------------------
I didn't realize that we now allow stepping with a '0' step. It should probably have been mentioned in the NEWS of 2.1.0?

Anyways, couple of bugs with that new feature:

    bn = 1 << 100
    bn.step(by: 0, to: bn).first(2) # => [bn, bn] ok
    bn.step(by: 0).first(2)         # => [bn.to_f, bn.to_f] not ok
    bn.step(by: 0, to: 0).first(2)  # => [] not ok

The corresponding `size` don't all work either:

    bn.step(by: 0) # => Float::INFINITY, ok
    bn.step(by: 0, to: bn).size # => ZeroDivisionError: divided by 0, should be infinity
    bn.step(by: 0, to: 0).size  # => same
    1.step(by:0, to: 42).size   # => same

My patch is almost finished.



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