Issue #8842 has been updated by mame (Yusuke Endoh).

File integer-with-range.pdf added


----------------------------------------
Bug #8842: Integer#[] with range
https://bugs.ruby-lang.org/issues/8842#change-41451

Author: mame (Yusuke Endoh)
Status: Assigned
Priority: Normal
Assignee: matz (Yukihiro Matsumoto)
Category: 
Target version: next minor
ruby -v: trunk
Backport: 1.9.3: UNKNOWN, 2.0.0: UNKNOWN


=begin
I propose to extend Integer#[] accepting a range.

    0b01001101[2, 4] == 0b0011
    0bHGFEDCBA[2, 4] == 0bFEDC

== Use case

I believe that everyone has written a code like this:

    if (n >> 2) & 0xf == 0x3
      ...
    end

because this is a very common idiom in C.
But it is less readable, writable, extendable and optimizable.

    if n[2, 4] == 0x3
      ...
    end

is much better in the all aspects.

== Corner cases

The current Integer#[] (and shift operators) handle an integer as "a bit array with infinity length";
it returns 0 for any negative index and an (extended) sign bit for any index greater than MSB.
We also can use this standard to define the spec for a range argument.
For example:

    15[-1, 42] #=> 30  (equivalent to (15 << 1) && (2 ** 42 - 1))
    15[3, 42]  #=>  1  (equivalent to (15 >> 3) && (2 ** 42 - 1))
    15[3..Float::INFINITY]  #=> 1  (equivalent to 15 >> 3)
    15[-3..Float::INFINITY] #=> 2  (equivalent to 1 << 3)

    -1[0..Float::INFINITY]   #=> -1
    -1[1..Float::INFINITY]   #=> -1
    -1[-1..Float::INFINITY]  #=> -2

    1[-Float::INFINITY..0] #=> failed to allocate memory
    2[-Float::INFINITY..0] #=> 0


Only tricky case that I thought of is a range (beg..end) whose "end" is smaller than "beg".
I think it should be handled as (beg..Float::INFINITY).

    15[-3..-4] #=> 2  (equivalent to 1 << 3)
    -1[0..-1] #=> -1
    -1[0..-2] #=> -1

What do you think?
=end



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