Asp Forum - inject's pathological case...

Randy R

3/30/2008 5:14:00 AM

I'm not sure how I feel about inject's pathological case. What do you
all think should happen in the following code?

[2].inject {|a, i| puts i}

I might have thought that the block would never be called and the method
would return 2.
If the block must be called, I might expect "i" to be nil.
What actually happens is that the block is called and "i" is the same
thing as "a". I did not expect this and I'm trying to figure out how this
is either consistent or useful. Can someone clue me in to this behaviour?
Thank you...

22 Answers

7stud --

3/30/2008 6:21:00 AM

Just Another Victim... wrote:
> I'm not sure how I feel about inject's pathological case. What do you
> all think should happen in the following code?
>
>
> [2].inject {|a, i| puts i}
>
>
> I might have thought that the block would never be called and the
> method
> would return 2.
> If the block must be called, I might expect "i" to be nil.

> What actually happens is that the block is called and "i" is the
> same
> thing as "a". I did not expect this and I'm trying to figure out how
> this
> is either consistent or useful. Can someone clue me in to this
> behaviour?

I'm not seeing that:

1) Two element array:

result = [2, 3].inject do |a, i|
puts "a:#{a}"
puts "i:#{i}"
puts "hello"
puts "--------"
end

puts "result=#{result}"
if result.nil?
puts 'yes'
end

--output:--
a:2
i:3
hello
--------
result=
yes

2) One element array:

result = [2].inject do |a, i|
puts "a:#{a}"
puts "i:#{i}"
puts "hello"
puts "--------"
end

puts "result=#{result}"
if result.nil?
puts 'yes'
end

--output:--
result=2
--
Posted via http://www.ruby-....

7stud --

3/30/2008 6:27:00 AM

Oh, yeah:

$ ruby -v
ruby 1.8.2 (2004-12-25) [universal-darwin8.0]

By the way, using inject() is inefficient--not to mention confusing.
You might as well pretend it doesn't exist.
--
Posted via http://www.ruby-....

David A. Black

3/30/2008 7:14:00 AM

Hi --

On Sun, 30 Mar 2008, Just Another Victim of the Ambient Morality wrote:

> I'm not sure how I feel about inject's pathological case. What do you
> all think should happen in the following code?
>
>
> [2].inject {|a, i| puts i}
>
>
> I might have thought that the block would never be called and the method
> would return 2.
> If the block must be called, I might expect "i" to be nil.
> What actually happens is that the block is called and "i" is the same
> thing as "a". I did not expect this and I'm trying to figure out how this
> is either consistent or useful. Can someone clue me in to this behaviour?
> Thank you...

Actually the block isn't being called:

irb(main):005:0> [2].inject {|a, i| puts "hi" }
=> 2
irb(main):006:0> [2].inject {}
=> 2

The value of the whole statement is 2, or, more generally, the first
(and only) element in the enumerable. I guess that's all it can do
since it doesn't have enough elements to call the block even once.

David

--
Rails training from David A. Black and Ruby Power and Light:
ADVANCING WITH RAILS April 14-17 New York City
INTRO TO RAILS June 9-12 Berlin
ADVANCING WITH RAILS June 16-19 Berlin
See http://www.r... for details and updates!

7stud --

3/30/2008 9:39:00 AM

David A. Black wrote:
>
> Actually the block isn't being called:
> I guess that's all it can do
> since it doesn't have enough elements to call the block even once.
>

To add to that, on p.456 of pickaxe2 it says that if you don't supply an
argument for the inject call, then the first element in the enumerable
becomes 'a', and it is not included in subsequent iteration.
Apparently, because there are no values to iterate over, the block does
not execute.

--
Posted via http://www.ruby-....

William James

3/30/2008 9:55:00 AM

On Mar 30, 12:27 am, 7stud -- <bbxx789_0...@yahoo.com> wrote:
> Oh, yeah:
>
> $ ruby -v
> ruby 1.8.2 (2004-12-25) [universal-darwin8.0]
>
> By the way, using inject() is inefficient--not to mention confusing.
> You might as well pretend it doesn't exist.

There is some truth to that.

def collect_repeats_inject list
return [] if [] == list
list[1..-1].inject([[ list.first ]]){|a,e|
if a[-1][0] == e
a[-1] << e
else
a << [e]
end
a
}.reject{|lst| lst.size < 2 }
end

def collect_repeats list
accum = [ [ list.shift ] ]
list.each{|e|
if accum[-1][0] == e
accum[-1] << e
else
accum << [e]
end }
accum.reject{|lst| lst.size < 2 }
end

p collect_repeats( %w(0 1 1 2 3 3 3 3 4 5 5 6) )
p collect_repeats( [] )
p collect_repeats_inject( %w(0 1 1 2 3 3 3 3 4 5 5 6) )
p collect_repeats_inject( [] )

the_list = %w(0 1 1 2 3 3 3 3 3 3 3 3 4 5 5 6 7 8 8 9 9 9)

t = Time.now
9999.times{ collect_repeats_inject( the_list )}
p Time.now - t
t = Time.now
9999.times{ collect_repeats( the_list )}
p Time.now - t

--- output ---
[["1", "1"], ["3", "3", "3", "3"], ["5", "5"]]
[]
[["1", "1"], ["3", "3", "3", "3"], ["5", "5"]]
[]
2.694
0.16

And the version without inject is shorter and
clearer.

William James

3/30/2008 10:04:00 AM

On Mar 30, 12:27 am, 7stud -- <bbxx789_0...@yahoo.com> wrote:
> Oh, yeah:
>
> $ ruby -v
> ruby 1.8.2 (2004-12-25) [universal-darwin8.0]
>
> By the way, using inject() is inefficient--not to mention confusing.
> You might as well pretend it doesn't exist.

There is some truth to that.

def collect_repeats_inject list
return [] if [] == list
list[1..-1].inject([[ list.first ]]){|a,e|
if a[-1][0] == e
a[-1] << e
else
a << [e]
end
a
}.reject{|lst| lst.size < 2 }
end

def collect_repeats list
return [] if [] == list
accum = [ [ list.first ] ]
list[1..-1].each{|e|
if accum[-1][0] == e
accum[-1] << e
else
accum << [e]
end }
accum.reject{|lst| lst.size < 2 }
end

p collect_repeats( %w(0 1 1 2 3 3 3 3 4 5 5 6) )
p collect_repeats( [] )
p collect_repeats_inject( %w(0 1 1 2 3 3 3 3 4 5 5 6) )
p collect_repeats_inject( [] )

the_list = %w(0 1 1 2 3 3 3 3 3 3 3 3 4 5 5 6 7 8 8 9 9 9)

t = Time.now
9999.times{ collect_repeats_inject( the_list )}
p Time.now - t
t = Time.now
9999.times{ collect_repeats( the_list )}
p Time.now - t

--- output ---
[["1", "1"], ["3", "3", "3", "3"], ["5", "5"]]
[]
[["1", "1"], ["3", "3", "3", "3"], ["5", "5"]]
[]
2.814
1.923

And the version without inject seems clearer.

Rick DeNatale

3/30/2008 1:34:00 PM

On 3/30/08, 7stud -- <bbxx789_05ss@yahoo.com> wrote:

> By the way, using inject() is inefficient--not to mention confusing.
> You might as well pretend it doesn't exist.

Or much better, understand that it's a way to reduce/fold the elements
of an enumerable into a single value.
http://en.wikipedia.org/wiki/Fold_%28higher-order_f...

Enumerable#inject is very useful, as long as you don't treat it as
Maslow's hammer and use it in inappropriate ways, such as giving it a
block like in the original posting to this thread.

--
Rick DeNatale

My blog on Ruby
http://talklikeaduck.denh...

Marc Heiler

3/30/2008 2:31:00 PM

> Understand that it's a way to reduce/fold the elements
> of an enumerable into a single value.

Not that this is important, but personally I never liked the name
inject
Though fold(l/r) is not much better either.

A peculiar thing is that the wikipedia page gives "also known variously
as " four different names as alternative/example.
--
Posted via http://www.ruby-....

Robert Dober

3/30/2008 2:33:00 PM

On Sun, Mar 30, 2008 at 3:34 PM, Rick DeNatale <rick.denatale@gmail.com> wrote:
> On 3/30/08, 7stud -- <bbxx789_05ss@yahoo.com> wrote:
>
> > By the way, using inject() is inefficient--not to mention confusing.
> > You might as well pretend it doesn't exist.
>
> Or much better, understand that it's a way to reduce/fold the elements
> of an enumerable into a single value.
> http://en.wikipedia.org/wiki/Fold_%28higher-order_f...
>
> Enumerable#inject is very useful, as long as you don't treat it as
> Maslow's hammer and use it in inappropriate ways, such as giving it a
> block like in the original posting to this thread.
>
> --
> Rick DeNatale
>
> My blog on Ruby
> http://talklikeaduck.denh...
>
>
Completely agree with Rick and furthermore you will not be able to
understand Ruby code written by others if you do not have a basic
understanding of inject.

I found this code in Ruby1.9 though

if result.size > 0 and result.inject(false) {|k,s| s or k}

that kind of code explains why inject has a bad reputation.

Cheers
Robert

--
http://ruby-smalltalk.blo...

---
Whereof one cannot speak, thereof one must be silent.
Ludwig Wittgenstein

Rick DeNatale

3/30/2008 3:17:00 PM

On Sun, Mar 30, 2008 at 10:30 AM, Marc Heiler <shevegen@linuxmail.org> wrote:
> > Understand that it's a way to reduce/fold the elements
> > of an enumerable into a single value.
>
> Not that this is important, but personally I never liked the name
> .inject
> Though fold(l/r) is not much better either.
>
> A peculiar thing is that the wikipedia page gives "also known variously
> as " four different names as alternative/example.

It's because the same thing has surfaced in several languages using
different terms, and those languages have in turn influenced others.

AFAIK, Matz correct me if I'm wrong, Ruby got the name inject from
Smalltalk, along with collect, select, detect and several other
methods. The other three I mention have alias, map for collect,
find_all for select, and find for detect, but inject is still just
inject in Ruby.

--
Rick DeNatale

My blog on Ruby
http://talklikeaduck.denh...

comp.lang.ruby

inject's pathological case...

Randy R

7stud --

7stud --

David A. Black

7stud --

William James

William James

Rick DeNatale

Marc Heiler

Robert Dober

Rick DeNatale

x Login to ForumsZone