- ruby-dev - ml.ruby-lang.org

[ruby-dev:52085] メール通知: 電子メールアカウントをアップグレードします。 Ref#UdnpS
by MailNotification＠ruby-lang.org 10 May '24

10 May '24

1 0

[ruby-dev:52084] [Ruby master Feature#5133] Array#unzip as an alias of Array#transpose
by matz (Yukihiro Matsumoto) 17 Apr '24

17 Apr '24

Issue #5133 has been updated by matz (Yukihiro Matsumoto). Status changed from Assigned to Rejected `unzip` is not really intuitive with Ruby's OO design. Unlike Haskell, Ruby does not have static type issues. Matz. ---------------------------------------- Feature #5133: Array#unzip as an alias of Array#transpose https://bugs.ruby-lang.org/issues/5133#change-107939 * Author: mrkn (Kenta Murata) * Status: Rejected * Assignee: mrkn (Kenta Murata) ---------------------------------------- Array#zip の逆は Array#transpose なんですけど、この対応関係が非常に分かり難いなと思いました。 Haskell には zip の逆をやる関数として unzip が用意されています。 unzip という名前は、「zip の逆をやりたい」と思ったときに (transpose よりは) 思い付きやすい名前だと思います。ということで Array#unzip を Array#transpose のエイリアスとして導入してはどうでしょう？以下パッチです: diff --git a/array.c b/array.c index 8caad66..dc411b7 100644 --- a/array.c +++ b/array.c @@ -4720,6 +4720,7 @@ Init_Array(void) rb_define_method(rb_cArray, "reject!", rb_ary_reject_bang, 0); rb_define_method(rb_cArray, "zip", rb_ary_zip, -1); rb_define_method(rb_cArray, "transpose", rb_ary_transpose, 0); + rb_define_alias(rb_cArray, "unzip", "transpose"); rb_define_method(rb_cArray, "replace", rb_ary_replace, 1); rb_define_method(rb_cArray, "clear", rb_ary_clear, 0); rb_define_method(rb_cArray, "fill", rb_ary_fill, -1); -- https://bugs.ruby-lang.org/

1 0

[ruby-dev:52082] [Ruby master Feature#5133] Array#unzip as an alias of Array#transpose
by duerst 22 Mar '24

22 Mar '24

Issue #5133 has been updated by duerst (Martin Dürst). In issue #20336, @matheusrich wrote: ``` * [Feature #5133] Array#unzip as an alias of Array#transpose * Seems a more friendly name for this method (easier if you don't have a strong math background) * It is nice that we can do an operation an reverse it with two similar-named methods: ```rb [1, 2, 3, 4].zip(["a", "b", "c", "d"], ["I", "II", "III", "IV"]) # => [[[1, "a"], "I"], [[2, "b"], "II"], [[3, "c"], "III"], [[4, "d"], "IV"]] [[[1, "a"], "I"], [[2, "b"], "II"], [[3, "c"], "III"], [[4, "d"], "IV"]].unzip # => [[1, 2, 3, 4], ["a", "b", "c", "d"], ["I", "II", "III", "IV"]] ``` It's already alluded in some of the comments above: `transpose` is it's own inverse, so if we define `unzip` as an alias for `transpose`, `unzip` will be its own inverse. The reason Haskell has `zip`/`unzip` as inverses is that `zip` in Haskell converts from two lists to a list of pairs, and `unzip` converts from a list of pairs to a pair of lists. In Haskell, `zip` and `unzip` are inverses (up to (un)currying). In Haskell, the arguments of `zip` are limited to two lists. In Ruby, it's one array as receiver and one or more arrays as arguments. For `transpose`, it's an array of arrays. Only the later is fully generic and thus can be made exactly invertible. ---------------------------------------- Feature #5133: Array#unzip as an alias of Array#transpose https://bugs.ruby-lang.org/issues/5133#change-107431 * Author: mrkn (Kenta Murata) * Status: Assigned * Assignee: mrkn (Kenta Murata) ---------------------------------------- Array#zip の逆は Array#transpose なんですけど、この対応関係が非常に分かり難いなと思いました。 Haskell には zip の逆をやる関数として unzip が用意されています。 unzip という名前は、「zip の逆をやりたい」と思ったときに (transpose よりは) 思い付きやすい名前だと思います。ということで Array#unzip を Array#transpose のエイリアスとして導入してはどうでしょう？以下パッチです: diff --git a/array.c b/array.c index 8caad66..dc411b7 100644 --- a/array.c +++ b/array.c @@ -4720,6 +4720,7 @@ Init_Array(void) rb_define_method(rb_cArray, "reject!", rb_ary_reject_bang, 0); rb_define_method(rb_cArray, "zip", rb_ary_zip, -1); rb_define_method(rb_cArray, "transpose", rb_ary_transpose, 0); + rb_define_alias(rb_cArray, "unzip", "transpose"); rb_define_method(rb_cArray, "replace", rb_ary_replace, 1); rb_define_method(rb_cArray, "clear", rb_ary_clear, 0); rb_define_method(rb_cArray, "fill", rb_ary_fill, -1); -- https://bugs.ruby-lang.org/

1 0

[ruby-dev:52080] [Ruby master Feature#6317] Range#cover?の引数としてRangeインスタンスを受けられるようにして欲しい
by matheusrich (Matheus Richard) 20 Mar '24

20 Mar '24

Issue #6317 has been updated by matheusrich (Matheus Richard). @mame @naruse This currently works on Ruby. Can we close this as completed? ---------------------------------------- Feature #6317: Range#cover?の引数としてRangeインスタンスを受けられるようにして欲しい https://bugs.ruby-lang.org/issues/6317#change-107335 * Author: masaakiaoyagi (Masaaki Aoyagi) * Status: Assigned * Assignee: matz (Yukihiro Matsumoto) ---------------------------------------- 青柳と申します。以下のように、Range#cover?の引数としてRangeインスタンスを受けられるようにして欲しいです。 (1..4).cover?(2..3) # => true (1..4).cover?(0..3) # => false 取り敢えず実装してみましたので、添付いたします。 ---Files-------------------------------- range_cover.diff (2.17 KB) -- https://bugs.ruby-lang.org/

1 0

[ruby-dev:52076] [Ruby master Bug#16297] calling undefined allocator by `Class.instance_method(:allocate)`
by jhawthorn (John Hawthorn) 01 Mar '24

01 Mar '24

Issue #16297 has been updated by jhawthorn (John Hawthorn). I'm not sure if there's a deeper problem this causes, but it is still possible after this patch to call the allocator on a Rational through defining an arbitrary `allocate` method. ``` ruby >> def Rational.allocate; end => :allocate >> Class.instance_method(:allocate).bind_call(Rational) => (0/1) ``` ---------------------------------------- Bug #16297: calling undefined allocator by `Class.instance_method(:allocate)` https://bugs.ruby-lang.org/issues/16297#change-107083 * Author: nobu (Nobuyoshi Nakada) * Status: Closed * Backport: 2.5: REQUIRED, 2.6: DONE ---------------------------------------- For instance, `Rational.allocate` is undefined. ```ruby Rational.allocate #=> undefined method `allocate' for Rational:Class (NoMethodError) ``` But it can be called by `Class.instance_method(:allocate)`. ```ruby Class.instance_method(:allocate).bind_call(Rational) #=> (0/1) Class.instance_method(:allocate).bind(Rational).call #=> (0/1) ``` These allocators are defined for Marshal, and undefined as methods. -- https://bugs.ruby-lang.org/

1 0

[ruby-dev:52068] [Ruby master Bug#10436] ruby -c and ripper inconsistency: m(&nil) {}
by yui-knk (Kaneko Yuichiro) 20 Feb '24

20 Feb '24

Issue #10436 has been updated by yui-knk (Kaneko Yuichiro). Status changed from Open to Closed This is solved by https://bugs.ruby-lang.org/issues/20257 (https://github.com/ruby/ruby/pull/9923) ---------------------------------------- Bug #10436: ruby -c and ripper inconsistency: m(&nil) {} https://bugs.ruby-lang.org/issues/10436#change-106914 * Author: akr (Akira Tanaka) * Status: Closed * Priority: Normal * ruby -v: ruby 2.2.0dev (2014-10-27 trunk 48168) [x86_64-linux] * Backport: 2.0.0: UNKNOWN, 2.1: UNKNOWN ---------------------------------------- m(&nil) {} というまちがったコードは ruby -c でちゃんとエラーになるのに、 Ripper.sexp ではならないようです。 ``` % cat z.rb m(&nil) {} % ruby -c z.rb z.rb:1: both block arg and actual block given % ruby -rripper -e 'p Ripper.sexp(STDIN.read)' < z.rb [:program, [[:method_add_block, [:method_add_arg, [:fcall, [:@ident, "m", [1, 0]]], [:arg_paren, [:args_add_block, [], [:var_ref, [:@kw, "nil", [1, 3]]]]]], [:brace_block, nil, [[:void_stmt]]]]]] % ruby -v ruby 2.2.0dev (2014-10-27 trunk 48168) [x86_64-linux] ``` ---Files-------------------------------- ripper-block_dup_check-10436.patch (2.2 KB) -- https://bugs.ruby-lang.org/

1 0

[ruby-dev:52066] [Ruby master Bug#20173] Backport 597955a, 8b65d15
by make_now_just (Hiroya Fujinami) 02 Feb '24

02 Feb '24

Issue #20173 has been reported by make_now_just (Hiroya Fujinami). ---------------------------------------- Bug #20173: Backport 597955a, 8b65d15 https://bugs.ruby-lang.org/issues/20173 * Author: make_now_just (Hiroya Fujinami) * Status: Open * Priority: Normal * Backport: 3.0: UNKNOWN, 3.1: UNKNOWN, 3.2: UNKNOWN, 3.3: UNKNOWN ---------------------------------------- please backport! -- https://bugs.ruby-lang.org/

2 1

[ruby-dev:52065] [Ruby master Feature#16142] Implement code_range in Proc and Method
by byroot (Jean Boussier) 09 Jan '24

09 Jan '24

Issue #16142 has been updated by byroot (Jean Boussier). > So what would be a good name? What about `position`? I think it goes well together with `location`. ---------------------------------------- Feature #16142: Implement code_range in Proc and Method https://bugs.ruby-lang.org/issues/16142#change-106130 * Author: okuramasafumi (Masafumi OKURA) * Status: Open * Priority: Normal ---------------------------------------- # Abstract Add a new method `code_range` as an alternative to `source_location` to Proc and Method # Background I'd like to get a body from a Proc in TraceLocation gem (https://github.com/yhirano55/trace_location) in order to add what's executed to the output. There's no way to do that in current Ruby implementation, so as an alternative, I considered getting source code location of a Proc. # Proposal I propose that `Proc#code_range` and `Method#code_range`. Other names can work as well, for example `Proc#source_region`. It returns an array containing filename as a first argument and position information as a second array. For example: `a_proc.position # => [(irb), [1, 5, 3, 25]]` # Implementation I've implemented a simpler version of this, see gist for more details. https://gist.github.com/okuramasafumi/ac90bbf04a1c13b7d67954c9c5e62553 Notice I use `code_location` from iseq struct. # Discussion One might say that we can simply add columns and end position to Proc#source_location. However, this can easily brake existing apps such as Pry. It's also possible that we add additional keyword argument to `Proc#source_location`, for instance: `a_proc.source_location(including_range: true)` This change can also break existing apps since in old Rubies this keyword argument cannot be accepted. Therefore, adding a new method is better in terms of backward compatibility. It might be better at readability as well. # Summary I propose an API to get code position of Proc and Method so that we can get body of them (especially of a Proc). -- https://bugs.ruby-lang.org/

1 0

[ruby-dev:52064] [Ruby master Feature#16142] Implement code_range in Proc and Method
by okuramasafumi (Masafumi OKURA) 09 Jan '24

09 Jan '24

Issue #16142 has been updated by okuramasafumi (Masafumi OKURA). Hi I noticed we didn't get any conclusion on this topic. I agree that method name should be something different. It follows the Matz's question: "What do we need from the method? line numbers? offset? whatever? " https://docs.ruby-lang.org/en/master/IO.html#method-c-read And we have `IO.read` method that accepts `length` and `offset` parameter. Since it's expected to be used with `IO.read`, it is convenient to return these two values. And Matz says > Making the return value as a specific class seems overkill. And I agree. So the return value could be either Array or Hash. I think Hash is better than Array in this case because the value such as `[19, 10]` doesn't make sense itself. So what would be a good name? The combination of length and offset doesn't have a specific name as far as I know. What about `location` if `Proc` has this method? The final version would look like this. ```ruby my_proc = proc do do_something end loc = my_proc.location p loc # => {length:10, offset:19} maybe wrong number p File.read(__FILE__, loc[:length], loc[:offset]) # => do_something ``` ---------------------------------------- Feature #16142: Implement code_range in Proc and Method https://bugs.ruby-lang.org/issues/16142#change-106122 * Author: okuramasafumi (Masafumi OKURA) * Status: Open * Priority: Normal ---------------------------------------- # Abstract Add a new method `code_range` as an alternative to `source_location` to Proc and Method # Background I'd like to get a body from a Proc in TraceLocation gem (https://github.com/yhirano55/trace_location) in order to add what's executed to the output. There's no way to do that in current Ruby implementation, so as an alternative, I considered getting source code location of a Proc. # Proposal I propose that `Proc#code_range` and `Method#code_range`. Other names can work as well, for example `Proc#source_region`. It returns an array containing filename as a first argument and position information as a second array. For example: `a_proc.position # => [(irb), [1, 5, 3, 25]]` # Implementation I've implemented a simpler version of this, see gist for more details. https://gist.github.com/okuramasafumi/ac90bbf04a1c13b7d67954c9c5e62553 Notice I use `code_location` from iseq struct. # Discussion One might say that we can simply add columns and end position to Proc#source_location. However, this can easily brake existing apps such as Pry. It's also possible that we add additional keyword argument to `Proc#source_location`, for instance: `a_proc.source_location(including_range: true)` This change can also break existing apps since in old Rubies this keyword argument cannot be accepted. Therefore, adding a new method is better in terms of backward compatibility. It might be better at readability as well. # Summary I propose an API to get code position of Proc and Method so that we can get body of them (especially of a Proc). -- https://bugs.ruby-lang.org/

1 0

[ruby-dev:52063] [Ruby master Bug#11183] Cumulative error on Complex::I ** 100000000000000000000000000000000
by kyanagi (Kouhei Yanagita) 16 Nov '23

16 Nov '23

Issue #11183 has been updated by kyanagi (Kouhei Yanagita). https://github.com/ruby/ruby/pull/8931 self の偏角が45度の整数倍であるような場合に特別扱いするようにしてみました。 ``` 1i ** (10 ** 100) #=> 1+0i 1i ** (10 ** 100 + 1) #=> 0+1i (1+1i) ** (10 ** 100) # warning: in a**b, b may be too big #=> (Infinity+0.0i) (1+1i) ** (10 ** 100 + 1) # warning: in a**b, b may be too big #=> (Infinity+Infinity*i) ``` * self が 1, -1, 1i, -1i の場合 * bignum 乗であっても正しい答えを返します。 * self が実数または純虚数で絶対値が1でない場合 * 偏角が 45 度の奇数倍である場合 * 小さい数（大きい数）が0（Infinity）となりますが、その前提のもとで正しい答えを返します。絶対値が 1 で偏角が 45 度の奇数倍である複素数は、実部・虚部（1/\sqrt{2}）をFloatで正確に表現することはできないため、偏角が45度の奇数倍である複素数オブジェクトの n 乗は、絶対値が1からずれていくことになります。なお、60 度など他の角度についてですが、偏角が 45 度の整数倍を除く有理数度である複素数は、Complex では正確には表現できません。（実部または虚部のいずれかが無理数になる）従って、Complex オブジェクトの偏角は 45 度の整数倍であるか無理数であるかになるため、偏角を n 倍することで n 乗を求める計算が誤差なしでできるのは、偏角が 45 度の整数倍の場合に限られることになります。 ---------------------------------------- Bug #11183: Cumulative error on Complex::I ** 100000000000000000000000000000000 https://bugs.ruby-lang.org/issues/11183#change-105340 * Author: ko1 (Koichi Sasada) * Status: Open * Priority: Normal * Backport: 2.0.0: UNKNOWN, 2.1: UNKNOWN, 2.2: UNKNOWN ---------------------------------------- ``` p Complex::I ** 100000000000000000000000000000000 ``` が、32bit 環境だと、 ``` ruby 2.3.0dev (2015-05-21 trunk 50502) [i386-mswin32_110] t.rb:1: warning: in a**b, b may be too big (-0.08483712490438944+1.5651333429325577e+32i) ``` こんな感じで、変な値が出ます。 ``` # ubuntu 64bit だと (0.9943722416631883-0.10594264962575697i) ``` できれば、こっちも 1+0i にぴたっとしてるといいのでしょうが。この問題に対する、うささんのコメント: > 真面目にコードを読むと、Complex#** は引数(指数)がFixnumでない場合は複素数をいったん極座標形式に変換してからrとθの双方をそれぞれ計算し、最後にf_complex_polarを使ってComplexオブジェクトにしてるのですが、θはFloatなので当然に誤差が蓄積し、という感じです。まあ、誤差が蓄積しなかったとしても、f_complex_polarの中でせっかくθが0°・90°・180°・270°のケースを特別扱いしようとしてるのにdouble値を==で比較してるのでまったく救われそうにない、という感じではありますが。 > θをn倍する計算に対して、nが整数である場合は360°は0°と同じという性質を利用して誤差を蓄積しない計算をすればだいぶマシになるとは思いますが、しかしめんどくさいっすね。 ---Files-------------------------------- my_complex.diff (2.09 KB) -- https://bugs.ruby-lang.org/

1 0