[ruby-core:46718] [ruby-trunk - Feature #6758] Object#sequence

Issue #6758 has been updated by ngoto (Naohisa Goto).


In biology, the term "sequence" has special meanings for DNA and protein, and I'm already using the method name "sequence" and its abbreviation "seq" in my libraries and tools (e.g. bioruby). So, I have objection to the name Object#sequence and #seq.

According to WikiPedia, "sequence" has several field-specific meanings other than biology.
http://en.wikipedia.org/wiki/Sequence_%28disambiguation%29

I prefer "iterate" or "recurrence".
----------------------------------------
Feature #6758: Object#sequence
https://bugs.ruby-lang.org/issues/6758#change-28385

Author: merborne (kyo endo)
Status: Open
Priority: Normal
Assignee: 
Category: 
Target version: 


=begin

== Object#sequence

Let me propose a new method ((*Object#sequence*)).

次のような実装の((*Object#sequence*))を提案します。

    class Object
      def sequence(init=true, &blk)
        x = self
        Enumerator.new do |y|
          y << x if init
          loop { y << (x = yield x) }
        end
      end
    end

((*sequence*)) generate a sequence by applying a block recursively to the receiver object. The result is wrapped with a Enumerator object, thus it is set under lazy evaluation.

((*sequence*))は、そのレシーバオブジェクトを初期値として、渡されたブロック（漸化式）を繰り返し適用してシーケンスを生成します。適用の結果はEnumeratorオブジェクトでラップされているので、遅延評価されます。

== Usage;

使い方を示します。

    1.sequence { |x| x + 2 }.take(10) # => [1, 3, 5, 7, 9, 11, 13, 15, 17, 19]
    
    3.sequence { |x| x * 2 }.take(10) # => [3, 6, 12, 24, 48, 96, 192, 384, 768, 1536]
    
    [0, 1].sequence { |a, b| [b, a + b] }.take(10).map(&:first) # => [0, 1, 1, 2, 3, 5, 8, 13, 21, 34]
    
    [0, 1, 1].sequence { |a, b, c| [b, c, a + b + c] }.take(10).map(&:first) # => [0, 1, 1, 2, 4, 7, 13, 24, 44, 81]
    
    # square root 5
    a = 5
    eps = 0.0001
    1.0.sequence { |x| (x + a/x) / 2.0 }
       .each_cons(2)
       .detect { |a, b| (a - b).abs < eps }[1] # => 2.236067977499978
    
    # Excel label
    'A'.sequence { |x| x.succ }.take(30) # => ["A", "B", "C", "D", "E", "F", "G", "H", "I", "J", "K", "L", "M", "N", "O", "P", "Q", "R", "S", "T", "U", "V", "W", "X", "Y", "Z", "AA", "AB", "AC", "AD"]
    
    # random boolean(avoid true-true sequence)
    true.sequence { |prev| prev ? false : [true, false].sample }.take(20) # => [true, false, true, false, true, false, true, false, true, false, true, false, true, false, true, false, false, false, true, false]
    
== Some background

== 提案の経緯

Let me explain some background of this request.

本件に関しては、ここに至る若干の経緯がありますので、併せて説明します。

1. I introduced this method as ((*Object#repeat*)) on my Japanese blog.

1. 私のブログにおいて、本メソッドを((*Object#repeat*))として紹介する記事を公開。

((<URL:http://melborne.github.com/2012/07/12/object-repeat-take-care-of-sequence/>))

2. Matz tweeted to the article.

    > "um.. the feature is attractive, but the name is.." 

2. 記事に対してMatzがつぶやく。

    > 「うーん、昨日としては魅力的だけど、名前がなあ。」

((<URL:https://twitter.com/yukihiro_matz/status/223790181113806848>))


3. I updated the article to propose ((*Object#repeat_apply*)) or ((*Object#repeat_call*)).

3. Object#repeat_apply, Object#repeat_callを提案するべく記事を更新。

4. Matz tweeted to the article.

    > @merborne more clear. but combining two verbs is wrong. I understand naming is difficult.

4. 記事に対してMatzがつぶやく。

    > @merborne なるほど「repeat_apply」か「repeat_call」ですか。repeat単体よりは誤解を受けにくいとは思いますが、repeatって動詞とapply/callって動詞の組み合わせは感心しませんね。名前って難しい。 

((<URL:https://twitter.com/yukihiro_matz/status/224105896110866432>))

5. Matz tweeted to the article.

    > @merborne I suggest some clue lies around a word "series"..

5. 記事に対してMatzがつぶやく。

    > @merborne 「級数/series」あたりに名前のヒントがありそうな。

((<URL:https://twitter.com/yukihiro_matz/status/224106160591081472>))

6. I tweeted to Matz.

6. 私もつぶやく

    > @yukihiro_matz you are right.. `repeated_apply` `repeated_call`?.. 


    > @yukihiro_matz たしかに.. repeated_apply repeated_callかな.. 

((<URL:https://twitter.com/merborne/status/224108387653259264>))

    > @yukihiro_matz clue! `series_by` ? 


    > @yukihiro_matz ヒント! series_by ? 

((<URL:https://twitter.com/merborne/status/224108809948377088>))

    > @yukihiro_matz `repeated` is adjective..^^; but I don't like `repeatedly_apply`. I thought once `series_by` is good, but it would be better a method is named based on its behavior, not on its function, I think. How about `repeat_by`?


    > @yukihiro_matz repeatedは形容詞でしたね^^; するとrepeatedly_applyですが、今ひとつです。series_byもいいと思ったんですが、できれば機能ではなく動作を言いたい思いがあります。そこでrepeat_byというのはどうでしょうか。

((<URL:https://twitter.com/merborne/status/224324670764220416>))


the conversation closed..

会話終了..

7. Ideas from other Rubyists

Some Japanese Rubyists tweeted or commented me on the name. Candidates are..

7. 他のRubyistの意見

名前に関し何人かのRubyistからアイディアをもらっています。リストアップします。

    iterate
    recur
    recurrence
    recur_with
    unfold
    sequence
    seq



Haskell and Scala have the same functionality with a name of ((*iterate*)).


なお、HaskellとScalaには同種の機能が、((*iterate*))という名前で実装されているそうです。


Also, @makotokuwata-san tweeted that ((*Kernel#seq*))(function style) or ((*Enumerator.seq()*))(a Class method) is better.


また、@makotokuwata氏よりKernel#seqのような関数形式か、Enumerator.seq()のようなクラスメソッドのほうがいいとの意見も頂いています。

    > @yukihiro_matz @merborne 初期値と漸化式から順列を作る機能なので、"sequence"か"seq"に1票。あとKernel#seq(initial,&blk)のほうが好きです。RT 「級数/series」あたりに名前のヒントがありそうな。 

((<URL:https://twitter.com/makotokuwata/status/225806204390227968>))



Thank you for your consideration.

以上、ご検討のほどよろしくお願い致します。


=end



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