ruby-core@ruby-lang.org archive (unofficial mirror)
 help / color / mirror / Atom feed
From: merch-redmine@jeremyevans.net
To: ruby-core@ruby-lang.org
Subject: [ruby-core:94909] [Ruby master Misc#16157] What is the correct and *portable* way to do generic delegation?
Date: Wed, 11 Sep 2019 21:25:30 +0000 (UTC)	[thread overview]
Message-ID: <redmine.journal-81522.20190911212529.f1d3979fab5b7bfb@ruby-lang.org> (raw)
In-Reply-To: redmine.issue-16157.20190909141042@ruby-lang.org

Issue #16157 has been updated by jeremyevans0 (Jeremy Evans).


After discussion with some other committers, I have an alternative approach in https://github.com/ruby/ruby/pull/2449.  This allows you to do the following:

```ruby
class Foo
  def foo(*args, &block)
    bar(*args, &block)
  end
  pass_keywords :foo if respond_to?(:pass_keywords, true)
end
```

Then if you call `foo` with keywords, keywords will be passed to `bar`.  This should allow for more backwards compatible behavior with older versions of Ruby.

Using a modified version of your example:

```ruby
class Proxy < BasicObject
  def initialize(target)
    @target = target
  end
  def method_missing(*a, &b)
    @target.send(*a, &b) rescue $!
  end
  pass_keywords(:method_missing) if respond_to?(:pass_keywords, true)
end

class Test
  def noarg() end
  def noarg_o(**o) o end
  def arg(a) a  end
  def arg_o(a, **o) [a,o] end
  def opt_arg(a=nil) a end
  def opt_arg_o(a=nil, **o) [a,o] end
  def args(*a) a end
  def args_o(*a, **o) [a, o] end
end

e = Object.new
def e.write(*args) $stdout.print("warn! ") end
$stderr = e

opt = {}
hash = {k:42}
proxy = Proxy.new(Test.new)
%i'noarg arg opt_arg args'.each do |m|
  [m, :"#{m}_o"].each do |meth|
    p ["#{meth}(opt)", proxy.send(meth, opt)]
    p ["#{meth}(hash)", proxy.send(meth, hash)]
    p ["#{meth}(**hash)", proxy.send(meth, **opt)]
    p ["#{meth}(**opt)", proxy.send(meth, **opt)]
    p ["#{meth}(hash, **opt)", proxy.send(meth, hash, **opt)]
    p ["#{meth}(hash, **hash)", proxy.send(meth, hash, **hash)]
  end
end
```

Here's a table describing the differences and warnings (differences in ArgumentError messages are ignored).  All differences are due to empty keyword splats not being passed as positional arguments in 2.7, and I would consider them at least undesired behavior in 2.6, if not an outright bug.

```
    Code                           2.6              2.7-pass_keywords
-----------------------------------------------------------------------------
  | noarg(opt)              | #<ArgE:(1:0)>   | #<ArgE:(1:0)>
  | noarg(hash)             | #<ArgE:(1:0)>   | #<ArgE:(1:0)>
 d| noarg(**opt)            | #<ArgE:(1:0)>   | nil
  | noarg(**hash)           | #<ArgE:(1:0)>   | #<ArgE:unknown keyword: :k>
  | noarg(hash, **opt)      | #<ArgE:(2:0)>   | #<ArgE:(1:0)>
  | noarg(hash, **hash)     | #<ArgE:(2:0)>   | #<ArgE:(1:0)>
w | noarg_o(opt)            | opt             | opt
w | noarg_o(hash)           | hash            | hash
  | noarg_o(**opt)          | opt             | opt
  | noarg_o(**hash)         | hash            | hash
  | noarg_o(hash, **opt)    | #<ArgE:(1:0)>   | #<ArgE:(1:0)>
  | noarg_o(hash, **hash)   | #<ArgE:(1:0)>   | #<ArgE:(1:0)>
  | arg(opt)                | opt             | opt
  | arg(hash)               | hash            | hash
w | arg(**opt)              | opt             | opt
  | arg(**hash)             | opt             | hash
 d| arg(hash, **opt)        | #<ArgE:(2:1)>   | hash
  | arg(hash, **hash)       | #<ArgE:(2:1)>   | #<ArgE:unknown keyword: :k>
  | arg_o(opt)              | [opt, opt]      | [opt, opt]
  | arg_o(hash)             | [hash, opt]     | [hash, opt]
w | arg_o(**opt)            | [opt, opt]      | [opt, opt]
w | arg_o(**hash)           | [hash, opt]     | [hash, opt]
  | arg_o(hash, **opt)      | [hash, opt]     | [hash, opt]
  | arg_o(hash, **hash)     | [hash, hash]    | [hash, hash]
  | opt_arg(opt)            | opt             | opt
  | opt_arg(hash)           | hash            | hash
 d| opt_arg(**opt)          | opt             | nil
  | opt_arg(**hash)         | hash            | hash
 d| opt_arg(hash, **opt)    | #<ArgE:(2:0..1)>| hash
  | opt_arg(hash, **hash)   | #<ArgE:(2:0..1)>| #<ArgE:unknown keyword: :k>
w | opt_arg_o(opt)          | [nil, opt]      | [nil, opt]
w | opt_arg_o(hash)         | [nil, hash]     | [nil, hash]
  | opt_arg_o(**opt)        | [nil, opt]      | [nil, opt]
  | opt_arg_o(**hash)       | [nil, hash]     | [nil, hash]
  | opt_arg_o(hash, **opt)  | [hash, opt]     | [hash, opt]
  | opt_arg_o(hash, **hash) | [hash, hash]    | [hash, hash]
  | args(opt)               | [opt]           | [opt]
  | args(hash)              | [hash]          | [hash]
 d| args(**opt)             | [opt]           | []
  | args(**hash)            | [hash]          | [hash]
 d| args(hash, **opt)       | [hash, opt]     | [hash]
  | args(hash, **hash)      | [hash, hash]    | [hash, hash]
w | args_o(opt)             | [[], opt]       | [[], opt]
w | args_o(hash)            | [[], hash]      | [[], hash]
  | args_o(**hash)          | [[], opt]       | [[], opt]
  | args_o(**opt)           | [[], opt]       | [[], opt]
  | args_o(hash, **opt)     | [[hash], opt]   | [[hash], opt]
  | args_o(hash, **hash)    | [[hash], hash]  | [[hash], hash]
```

Every time a warning is issued, behavior is the same in 2.7 as in 2.6. For each warning, here will be the behavior in 3.0:

```
    Code                 2.6 & 2.7           3.0
-------------------------------------------------------
w | noarg_o(opt)    | opt             | #<ArgE:(1:0)>
w | noarg_o(hash)   | hash            | #<ArgE:(1:0)>
w | arg(**opt)      | opt             | #<ArgE:(0:1)>
w | arg_o(**opt)    | [opt, opt]      | #<ArgE:(0:1)>
w | arg_o(**hash)   | [hash, opt]     | #<ArgE:(0:1)>
w | opt_arg_o(opt)  | [nil, opt]      | [opt, opt]
w | opt_arg_o(hash) | [nil, hash]     | [hash, opt]
w | args_o(opt)     | [[], opt]       | [[opt], opt]
w | args_o(hash)    | [[], hash]      | [[hash], opt]
```

----------------------------------------
Misc #16157: What is the correct and *portable* way to do generic delegation?
https://bugs.ruby-lang.org/issues/16157#change-81522

* Author: Dan0042 (Daniel DeLorme)
* Status: Open
* Priority: Normal
* Assignee: 
----------------------------------------
With the keyword argument changes in 2.7 we must now specify keyword arguments explicitly when doing generic delegation. But this change is not compatible with 2.6, where it adds an empty hash to the argument list of methods that do not need/accept keyword arguments.

To illustrate the problem:

```ruby
class ProxyWithoutKW < BasicObject
  def initialize(target)
    @target = target
  end
  def method_missing(*a, &b)
    @target.send(*a, &b)
  end
end

class ProxyWithKW < BasicObject
  def initialize(target)
    @target = target
  end
  def method_missing(*a, **o, &b)
    @target.send(*a, **o, &b)
  end
end

class Test
  def args(*a)   a  end
  def arg(a)     a  end
  def opts(**o)  o  end
end
                                          # 2.6        2.7              3.0
ProxyWithoutKW.new(Test.new).args(42)     # [42]       [42]             [42]        ok
ProxyWithoutKW.new(Test.new).arg(42)      # 42         42               42          ok
ProxyWithoutKW.new(Test.new).opts(k: 42)  # {:k=>42}   {:k=>42} +warn   [{:k=>42}]  incompatible with >= 2.7
ProxyWithKW.new(Test.new).args(42)        # [42, {}]   [42]             [42]        incompatible with <= 2.6
ProxyWithKW.new(Test.new).arg(42)         # error      42               42          incompatible with <= 2.6
ProxyWithKW.new(Test.new).opts(k: 42)     # {:k=>42}   {:k=>42} +warn   {:k=>42}    must ignore warning? cannot use pass_positional_hash in 2.6
```

I don't know how to solve this, so I'm asking for the **official** correct way to write portable delegation code. And by **portable** I mean code that can be used in gems that target ruby 2.6 and above.




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

  parent reply	other threads:[~2019-09-11 21:26 UTC|newest]

Thread overview: 25+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
     [not found] <redmine.issue-16157.20190909141042@ruby-lang.org>
2019-09-09 14:10 ` [ruby-core:94860] [Ruby master Misc#16157] What is the correct and *portable* way to do generic delegation? daniel
2019-09-09 14:22 ` [ruby-core:94861] " merch-redmine
2019-09-09 17:09 ` [ruby-core:94865] " daniel
2019-09-09 19:36 ` [ruby-core:94872] " merch-redmine
2019-09-10 14:41 ` [ruby-core:94888] " daniel
2019-09-10 16:37 ` [ruby-core:94889] " merch-redmine
2019-09-10 19:14 ` [ruby-core:94894] " daniel
2019-09-10 20:09 ` [ruby-core:94895] " merch-redmine
2019-09-11 15:36 ` [ruby-core:94904] " daniel
2019-09-11 21:25 ` merch-redmine [this message]
2019-09-11 23:20 ` [ruby-core:94912] " merch-redmine
2019-09-12  2:05 ` [ruby-core:94916] " daniel
2019-09-12 12:53 ` [ruby-core:94922] " daniel
2019-09-12 15:40 ` [ruby-core:94923] " merch-redmine
2019-09-12 19:57 ` [ruby-core:94924] " daniel
2019-09-13 14:55 ` [ruby-core:94927] " daniel
2019-09-13 15:15 ` [ruby-core:94928] " merch-redmine
2019-09-13 17:00 ` [ruby-core:94931] " daniel
2019-09-13 17:54 ` [ruby-core:94933] " merch-redmine
2019-10-14 18:10 ` [ruby-core:95317] " merch-redmine
2019-10-15  5:37 ` [ruby-core:95326] " eregontp
2019-10-15  7:23 ` [ruby-core:95328] " eregontp
2019-10-15 15:16 ` [ruby-core:95337] " merch-redmine
2019-10-15 15:49 ` [ruby-core:95339] " eregontp
2019-10-15 16:28 ` [ruby-core:95341] " merch-redmine

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-list from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

  List information: https://www.ruby-lang.org/en/community/mailing-lists/

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=redmine.journal-81522.20190911212529.f1d3979fab5b7bfb@ruby-lang.org \
    --to=ruby-core@ruby-lang.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).