Feature #6353: caller-side **

• Status: Closed
• Assignee: nobu (Nobuyoshi Nakada)
• Target version: 2.0.0

[mame (Yusuke Endoh)]

Relating to keyword argument (#5474), there are some requests for
caller-side **. [ruby-core:40518]

def foo(k1: 1, k2: 2)
p [k1, k2]
end
h = {k2: "bar"}
foo(k1: "foo", **h) # <== here
#=> ["foo", "bar"]

Marc-Andre explained the use case [ruby-core:41772], and matz agreed with
this feature. [ruby-core:41818]

However, it conflicts with power expression when parens are omitted:

foo **h # foo(h)? or foo.send("", h)?

Which should it be interpreted?

Anyway, I have no idea to avoid yacc conflict.
Nobu, could you please try to implement it?

--
Yusuke Endoh [e-mail]

[matz (Yukihiro Matsumoto)]

Hi,

In message "Re: [ruby-core:44591] [ruby-trunk - Feature #6353][Assigned] caller-side **"
on Tue, 24 Apr 2012 22:04:03 +0900, "mame (Yusuke Endoh)" [e-mail] writes:

|However, it conflicts with power expression when parens are omitted:
|
| foo **h # foo(h)? or foo.send("", h)?
|
|Which should it be interpreted?

It should be interpreted as *, i.e.

foo a # foo(a)
foo * a # foo.(a)
fooa # foo.*(a)

thus

foo h # foo(h)
foo ** h # foo.(h)
fooh # foo.**(h)

						matz.

[nobu (Nobuyoshi Nakada)]

=begin
Almost implemented, and another question.

What should happen in this case?

def foo(k1: 1)
p k1
end
h = {k1: "bar"}
foo(k1: "foo", **h) # <== conflict

(({k1})) will be (({"foo"})), or (({"bar"}))?
Or an exception should be raised?
=end

[nobu (Nobuyoshi Nakada)]

• Status changed from Assigned to Closed
• % Done changed from 0 to 100

This issue was solved with changeset r35489.
Yusuke, thank you for reporting this issue.
Your contribution to Ruby is greatly appreciated.
May Ruby be with you.

---

• parse.y (assoc, parser_yylex): add syntax to splat keyword hash.
  [ruby-core:44591][Feature #6353]
• compile.c (compile_array_): generate keyword splat insns.
• vm.c (m_core_hash_merge_kwd): merge keyword hash into intermediate
  hash. leftward argument is prior currently.