[ ruby-Bugs-9488 ] Date module, step method, infinite loop if +step+ is 0 ?

Bugs item #9488, was opened at 2007-03-23 04:02
You can respond by visiting: 
http://rubyforge.org/tracker/?func=detail&atid=1698&aid=9488&group_id=426

Category: Standard Library
Group: None
>Status: Closed
>Resolution: Rejected
Priority: 3
Submitted By: Colin Bartlett (colinb2)
Assigned to: Nobody (None)
Summary: Date module, step method, infinite loop if +step+ is 0 ?

Initial Comment:
My apologies in advance if this is known about.
I have looked at the outstanding "bugs" list,
and couldn't find anything.

I'm using the standard Ruby Installer for Windows
Ruby Version 1.8.5, Installer Version 185-21
ruby --version  ->  ruby 1.8.5 (2006-08-25) [i386-mswin32]

code of the step method in Tadayoshi Funaba's Date module
  # Step the current date forward +step+ days at a
  # time (or backward, if +step+ is negative) until
  # we reach +limit+ (inclusive), yielding the resultant
  # date at each step.
def step(limit, step)  # :yield: date
  da = self
  op = [:-,:<=,:>=][step<=>0]
  while da.__send__(op, limit)
    yield da
    da += step
  end
  self
end

Comment: the step method seems to go into an infinite loop
if +step+ is 0, continually yielding the start date (self).
I didn't instantly understand the "op = " and "while" lines,
but I think I understand them now, except if +step+ is 0
then the while expression becomes  ( da - limit ),
which is always going to be True because even if it's 0
it will be True? Is that why the infinite loop is happening?

Is an infinite loop the intended behaviour if +step+ is 0 ?
Or is the intention for people to rely on testing 
not to use a +step+ of 0? (That's not intended 
to be sarcastic: I don't know enough to be sure
the behaviour is not as intended.)

In the Integer step method a +step+ of 0 raises an exception,
which seems a better way to handle it? (Maybe have an option
of yielding (only once) the start date if +step+ is 0 ?
On reflection I think I still prefer the Integer step behaviour.)


 # The following is code to show the different behaviour
 # of Integer step and Date step when +step+ is 0.

$\ = "  " ; $, = "  "  # helps print output legibly on single lines

puts "\n ** Integer step example: exception raised if step = 0"
500.step( 505, 2 )  { | nxt |  print nxt }  # ->  500  502  504
50.step( 45, -2 )  { | nxt |  print nxt }  # ->  50  48  46
begin
  5.step( 5, 0 )  { | nxt |  puts nxt }
rescue 
  puts ' * exception raised: message = ' + $!.message
end  # ->   * exception raised: message = step can't be 0

require 'date'
c = Date.new( 2008, 1, 31 )
puts "\n\n ** Date step example:" + 
          " no error if step = 0, but an infinite loop?"
c.step( c + 3, 2 )   { | nxt |  print nxt }
                   # ->  2008-01-31  2008-02-02
c.step( c - 3, -2 )  { | nxt |  print nxt }
                   # ->  2008-01-31  2008-01-29
puts
count = 0
c.step( c + 3, 0 ) { | nxt |
                     puts count.to_s + ": " + nxt.to_s
                     count += 1
                     break if count > 10
                   }  # -> "infinite" loop of 2008-01-31 ?


----------------------------------------------------------------------

>Comment By: Ryan Davis (zenspider)
Date: 2007-03-23 04:13

Message:
Infinite loop with step 0 makes sense to me. It'll take me forever to get somewhere if I never move. If what you're really trying to say is that you think Date.step should raise if step == 0, then please be more clear and say that in the bug title (preferably in a new bug).

----------------------------------------------------------------------

Comment By: Colin Bartlett (colinb2)
Date: 2007-03-23 04:10

Message:
My original example used
  c.step( c, 0 )
but for the "bug" report I changed it to
  c.step( c + 3, 0 )
without testing it. (A bad idea.) If using
  c.step( c + 3, 0 )
doesn't show the Date step behaviour if +step+ is 0,
then please try my original example.

----------------------------------------------------------------------

You can respond by visiting: 
http://rubyforge.org/tracker/?func=detail&atid=1698&aid=9488&group_id=426