Bug #4451: Date#step does not raise ArgumentError for 0 step

• Status: Rejected
• Assignee: tadf (tadayoshi funaba)
• Target version: -
• ruby -v: ruby 1.9.3dev (2011-02-16 trunk 30888) [x86_64-openbsd4.9]

[jeremyevans0 (Jeremy Evans)]

=begin
Numeric#step:

$ ruby -e '1.step(2, 0){}'
-e:1:in step': step can't be 0 (ArgumentError) from -e:1:in '

Range#step:

$ ruby -r date -e '(0..1).step(0){}'
-e:1:in step': step can't be 0 (ArgumentError) from -e:1:in '

Date#step:

ruby -r date -e 'Date.today.step(Date.today, 0){}'

infinite loop

There is already a check for this in date.rb, but it is commented out. It was added in r13408, over 3 years ago. The lack of this check can cause a denial of service if the step value is controlled by the attacker. The attached patch enables the check and adds a test for it.
=end

[tadf (tadayoshi funaba)]

• Priority changed from Normal to 3

=begin
you are not right.
the intention of comment outed code is different.

i have a doubt about Numeric's one.
firstly, i think the current behaviour of date's one is not wrong.
however, i believe it should raise error soon if it should be an error.
but Numeric's one is not.

a = 1.step(1,0)
#=> #<Enumerator: 1:step(1, 0)>
a.each{}
ArgumentError: step can't be 0
from (irb):3:in step' from (irb):3:in each'
from (irb):3
from ./bin/irb:12:in `'

i once asked and proposed this.
however, i don't still get any answers.
so the change has been frozen.

i don't understand why Numeric#step delay the raise.
=end

[shyouhei (Shyouhei Urabe)]

• Status changed from Open to Assigned

[tadf (tadayoshi funaba)]

• Status changed from Assigned to Rejected