Bug #10653: do-end block in ternary operator is syntax error

• Status: Closed
• Assignee: matz (Yukihiro Matsumoto)
• Target version: -
• ruby -v: ruby 2.2.0p0 (2014-12-25 revision 49005) [x86_64-darwin14]
• Backport: 2.0.0: DONTNEED, 2.1: DONTNEED, 2.2: DONE

[yancya (Shinta Koyanagi)]

• For example

ruby -v -e 'true ? 1.tap do |n| p n end : 0'

• Result by 2.1.5

ruby 2.1.5p273 (2014-11-13 revision 48405) [x86_64-darwin14.0]
1

• Result by 2.2.0

ruby 2.2.0p0 (2014-12-25 revision 49005) [x86_64-darwin14]
-e:1: syntax error, unexpected keyword_do_cond, expecting ':'
true ? 1.tap do |n| p n end : 0
               ^
-e:1: syntax error, unexpected tIDENTIFIER, expecting keyword_do or '{' or '('
true ? 1.tap do |n| p n end : 0
                       ^

[nobu (Nobuyoshi Nakada)]

• Related to Feature #4276: Allow use of quotes in symbol syntactic sugar for hashes added

[nobu (Nobuyoshi Nakada)]

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

[nobu (Nobuyoshi Nakada)]

• Status changed from Open to Closed

Applied in changeset r51617.

---

parse.y: fix block after conditional

• parse.y: fix syntax error at do-block after a conditional
  operator. separate label-allowed and after-a-label states from
  others as bit flags. [ruby-dev:48790] [Bug #10653]

[nobu (Nobuyoshi Nakada)]

• Related to Bug #11456: Hash リテラル中で rescue をもちいると SyntaxError になる added

[nagachika (Tomoyuki Chikanaga)]

After backporting r50409 and r51617, TestSyntax#test_block_after_cond failed.

[nagachika (Tomoyuki Chikanaga)]

r51616 should be backported too?

[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

r50409, r51616, r51617 and r51624 were backported into ruby_2_2 branch at r52046.
With r50409 a minor incompatibility is introduced ({ %w"key": 0 } cause SyntaxError).
I'm not confident the incompatibility is essential part of this fixes, but the accurate surgery is required to eliminate the incompatibility. I gave up.

[whitequark (whitequark *)]

I was implementing this in my Ruby parser and I found a much simpler solution than what you've implemented here: https://github.com/whitequark/parser/commit/113442eff1f9c6f7e7205ec0dac8cb858b3b0fc7

In short, instead of explicitly enumerating all contexts where a label is valid, I only disallow quoted label at EXPR_VALUE using an additional flag. It works just as well. It also would not exhibit the bug with %w during backport.

Might I suggest a simplification to parse.y?

[nobu (Nobuyoshi Nakada)]

Currently it's a bit flag in parse.y too.