Bug #11501: About priority of a hash element when using {**hash1, **hash2} literal

• Status: Closed
• Assignee: nagachika (Tomoyuki Chikanaga)
• Target version: -
• ruby -v: ruby 2.1.3p242 (2014-09-19 revision 47630) [x86_64-darwin13.0]
• Backport: 2.0.0: DONTNEED, 2.1: DONTNEED, 2.2: DONE

[studio3104 (Satoshi SUZUKI)]

When creating a new hash using { **hash1, **hash2 } literal,
there are different behaviors between 2.2.0 or later and 2.1.7 or earlier like below.

• ruby 2.2.0p0 (2014-12-25 revision 49005) [x86_64-darwin13]
• ruby 2.1.3p242 (2014-09-19 revision 47630) [x86_64-darwin13.0]

[1] pry(main)> hash1 = { duplicate_key: 1, key: 1 }
=> {:duplicate_key=>1, :key=>1}
[2] pry(main)> hash2 = { duplicate_key: 3 }
=> {:duplicate_key=>3}
[3] pry(main)> { **hash1, **hash2 }
=> {:duplicate_key=>3, :key=>1}
[4] pry(main)>

• ruby 2.1.7p400 (2015-08-18 revision 51632) [x86_64-darwin13.0]

[1] pry(main)> hash1 = { duplicate_key: 1, key: 1 }
=> {:duplicate_key=>1, :key=>1}
[2] pry(main)> hash2 = { duplicate_key: 3 }
=> {:duplicate_key=>3}
[3] pry(main)> { **hash1, **hash2 }
=> {:duplicate_key=>1, :key=>1}
[4] pry(main)>

It seems that the behavior has changed from 2.2.0, however there is no mention in any documents as far as I searched.

• https://github.com/ruby/ruby/blob/v2_2_0/NEWS
• https://www.ruby-lang.org/en/news/2014/12/25/ruby-2-2-0-released/

Is this intended change?
Please check this and give me a opinion.
Warm Regards.

[studio3104 (Satoshi SUZUKI)]

• Description updated (diff)

[studio3104 (Satoshi SUZUKI)]

• Description updated (diff)

[studio3104 (Satoshi SUZUKI)]

• Description updated (diff)

[nagachika (Tomoyuki Chikanaga)]

• Status changed from Open to Assigned
• Assignee set to matz (Yukihiro Matsumoto)

Hello, satoshi san.
Thank you for your report.

I've checked the behaviors on each versions from 2.1.0-2.1.7, 2.2.0-2.2.4, and trunk(2.3-head).

script: hashes.rb

a = {a: 1, b: 2}
b = {a: 42}
p({ **a, **b })

results

→ for v in 2.1.0 2.1.1 2.1.2 2.1.3 2.1.4 2.1.5 2.1.6 2.1.7 2.2.0 2.2.1 2.2.2 2.2.3 2.3.0-dev; do rbenv shell $v; ruby -v hashes.rb; done
ruby 2.1.0p0 (2013-12-25 revision 44422) [x86_64-darwin13.0]
{:a=>1, :b=>2}
ruby 2.1.1p76 (2014-02-24 revision 45161) [x86_64-darwin13.0]
{:a=>1, :b=>2}
ruby 2.1.2p95 (2014-05-08 revision 45877) [x86_64-darwin13.0]
{:a=>1, :b=>2}
ruby 2.1.3p242 (2014-09-19 revision 47630) [x86_64-darwin13.0]
{:a=>1, :b=>2}
ruby 2.1.4p265 (2014-10-27 revision 48166) [x86_64-darwin13.0]
{:a=>1, :b=>2}
ruby 2.1.5p273 (2014-11-13 revision 48405) [x86_64-darwin14.0]
{:a=>1, :b=>2}
ruby 2.1.6p336 (2015-04-13 revision 50298) [x86_64-darwin14.0]
{:a=>1, :b=>2}
ruby 2.1.7p400 (2015-08-18 revision 51632) [x86_64-darwin14.0]
{:a=>1, :b=>2}
ruby 2.2.0p0 (2014-12-25 revision 49005) [x86_64-darwin13]
{:a=>42, :b=>2}
ruby 2.2.1p85 (2015-02-26 revision 49769) [x86_64-darwin14]
{:a=>42, :b=>2}
ruby 2.2.2p95 (2015-04-13 revision 50295) [x86_64-darwin14]
{:a=>42, :b=>2}
ruby 2.2.3p173 (2015-08-18 revision 51636) [x86_64-darwin14]
{:a=>42, :b=>2}
ruby 2.3.0dev (2015-10-07 trunk 52068) [x86_64-darwin14]
{:a=>42, :b=>2}

It seems that the behavior was changed between 2.1 <-> 2.2 series.
I don't known its intended or not.
In my personal opinion, I feel the latter (behavior of 2.2, 2.3) is reasonable because it's consistent with Hash#merge.

If matz admit this change, I'd like to add "Hash's incompatible changes" section in NEWS file on ruby_2_2 branch.

[wanabe (_ wanabe)]

git bisect shows the change was introduced by #10315.
I guess it is intended.

[wanabe (_ wanabe)]

• Related to Bug #10315: Override policy for duplicated keywords added

[wanabe (_ wanabe)]

Oops, many revisions references #10315.
Sorry, r47878 is the revision.

[nagachika (Tomoyuki Chikanaga)]

• Assignee changed from matz (Yukihiro Matsumoto) to nagachika (Tomoyuki Chikanaga)
• Backport changed from 2.0.0: UNKNOWN, 2.1: UNKNOWN, 2.2: UNKNOWN to 2.0.0: DONTNEED, 2.1: DONTNEED, 2.2: REQUIRED

wanabe san, thank you for your investigation!

According to #10315, It's intentional change.
I'll add comment for it in NEWS in ruby_2_2 and doc/NEWS-2.2.0 in trunk.

[nagachika (Tomoyuki Chikanaga)]

• Status changed from Assigned to Closed

Applied in changeset r53050.

---

• doc/NEWS-0.2.2: add description about incompatible change in Hash
  duplicated key overriding policy. [Bug #10315] [Bug #11501]

[nagachika (Tomoyuki Chikanaga)]

• Backport changed from 2.0.0: DONTNEED, 2.1: DONTNEED, 2.2: REQUIRED to 2.0.0: DONTNEED, 2.1: DONTNEED, 2.2: DONE

Add description to NEWS about this on ruby_2_2 branch (I committed in doc/NEWS-2.2.0 on trunk and backport it) at r53051.