From: "Guy N. Hurst" <gnhurst@...>
Date: 2001-05-02T01:36:17+09:00
Subject: [ruby-talk:14502] Re: RCR: Array#insert


Yukihiro Matsumoto wrote:
> 
> Hi,
> 
> In message "[ruby-talk:14486] Re: RCR: Array#insert"
>     on 01/05/01, "Guy N. Hurst" <gnhurst@hurstlinks.com> writes:
> 
> |What about his alternate version that handled negative indices
> |differently? [ruby-talk:14371] further down.
> 
> Hmm, there's tradeoff between inconsistency (to []= at least) and
> handiness.  Let me ponder for a while.  Discussions are welcome.


We did discuss this before also, in the thread starting at [ruby-talk:5692],
http://blade.nagaokaut.ac.jp/cgi-bin/scat.rb/ruby/ruby-talk/5692,
and Weirich's post [ruby-talk:5700] cleared it up, I think:

>>>
Don't think of it as "insert before" or "insert after".  Think of it
as "insert at".  After the assertion, the inserted value will be *at*
the given index.

E.g.

    After ...
        a.insert (index, value)
    then ...
        a[index] == value

whether index is positive or negative.
<<<

Currently, I consider the current approach as inconsistent.

Here's one aspect of it:

a=[1,2,3]
a[0,0] = "a"	# ["a",1,2,3]
a[1,0] = "b"	# ["a","b",1,2,3]
a[-1,0] = "c"	# ["a","b",1,2,"c",3]

One might think that they are inserting "at" a 
particular location, but when using negative
indices, realize it was always inserting "before".

but
b=[1,2,3,4]
b[0] = "a"	# ["a",2,3,4]
b[1] = "b"	# ["a","b",3,4,]
b[-1] = "c"	# ["a","b",3,"c"]

Here, we are replacing "at" the given index.




Aleksi brought it up again in [ruby-talk:5830],
http://blade.nagaokaut.ac.jp/cgi-bin/scat.rb/ruby/ruby-talk/5830.

Dave saw no reason to change []= (citing availability of #push),
nor for even adding #insert. 
(And you reminded me that there is #concat, and I also remember <<)

My point is that if we have #insert, why not make it able to
insert at *any* place in the array? But then, I also think
[]= should act the same way, which perhaps would break older
scripts relying on [-1,0]= to insert at [-2] ?

But consider that people have already implemented #insert
in Ruby. By making it part of Ruby, it will become "standard",
so I propose that it should handle negative indices properly,
regardless of how [pos,0]= works.

1) because it will help the transition of fixing [pos,0]= as 
people start to prefer using #insert over [pos,0]=
2) because it will eliminate the need for us to override
a 'wrong' #insert with our own version of #insert ;-)
3) would support using single function for inserting
anywhere, reducing code 


About reducing code....
Why should I always do this in Ruby:

a=[1,2,3]
val="value"
print "enter position to insert value:"
pos = gets.to_i
if pos == -1
  a.push(val)
elsif pos < -1 
  a.insert(pos+1,val)
else 
  a.insert(pos, val)
end

When I can do this:

a=[1,2,3]
val="value"
print "enter position to insert value:"
pos = gets.to_i
a.insert(pos,val)
 
????

Thank you for your time.


Guy N. Hurst

-- 
HurstLinks Web Development    http://www.hurstlinks.com/
Norfolk, VA  23510            (757)623-9688 FAX 623-0433
PHP/MySQL - Ruby/Perl - HTML/Javascript