JRuby

'obj !~ thing' is not the same as obj.send(:!~, thing)

Details

  • Type: Bug Bug
  • Status: Resolved Resolved
  • Priority: Major Major
  • Resolution: Fixed
  • Affects Version/s: JRuby 1.6.7
  • Fix Version/s: JRuby 1.7.0.pre1
  • Component/s: Parser, Ruby 1.9.2
  • Labels:
    None
  • Number of attachments :
    0

Description

I noticed this while working on some custom library code to implement '!~'

Patch to add this to RubySpec, which currently fails in JRuby 1.6.7: https://github.com/jruby/rubyspec/pull/2

irb session showing the failure:

% JRUBY_OPTS=--1.9 irb                      
>> JRUBY_VERSION + " " +  JRUBY_REVISION
=> "1.6.7 3e82bc8"
>> class Foo; def !~(val); return val; end; end
=> nil
>> Foo.new.send(:"!~", /bar/)
=> /bar/
>> Foo.new !~ /bar/
=> true

In the above example, 'Foo.new !~ /bar/' should have returned /bar/, not true.

Activity

Ascending order - Click to sort in descending order
Hide
Permalink
Charles Oliver Nutter added a comment -

Appears to be a parsing problem. It's parsing as in 1.8, as !(a =~ b):

system ~/projects/jruby $ ast -e 'a =~ b'
AST:
RootNode 0
  NewlineNode 0
    CallOneArgNode:=~ 0
      VCallNode:a 0
      ArrayNode 0
        VCallNode:b 0


system ~/projects/jruby $ ast -e 'a !~ b'
AST:
RootNode 0
  NewlineNode 0
    NotNode 0
      CallOneArgNode:=~ 0
        VCallNode:a 0
        ArrayNode 0
          VCallNode:b 0


system ~/projects/jruby $ jruby --1.8 -S ast -e 'a !~ b'
AST:
RootNode 0
  NewlineNode 0
    NotNode 0
      CallOneArgNode:=~ 0
        VCallNode:a 0
        ArrayNode 0
          VCallNode:b 0

It should parse as a call to :!~.

Show
Charles Oliver Nutter added a comment - Appears to be a parsing problem. It's parsing as in 1.8, as !(a =~ b): system ~/projects/jruby $ ast -e 'a =~ b' AST: RootNode 0 NewlineNode 0 CallOneArgNode:=~ 0 VCallNode:a 0 ArrayNode 0 VCallNode:b 0 system ~/projects/jruby $ ast -e 'a !~ b' AST: RootNode 0 NewlineNode 0 NotNode 0 CallOneArgNode:=~ 0 VCallNode:a 0 ArrayNode 0 VCallNode:b 0 system ~/projects/jruby $ jruby --1.8 -S ast -e 'a !~ b' AST: RootNode 0 NewlineNode 0 NotNode 0 CallOneArgNode:=~ 0 VCallNode:a 0 ArrayNode 0 VCallNode:b 0 It should parse as a call to :!~.
Hide
Permalink
Hiro Asari added a comment -

I think you meant obj !~ thing.

Show
Hiro Asari added a comment - I think you meant obj !~ thing .
Hide
Permalink
Charles Oliver Nutter added a comment -

Here's a hacky parser fix, but I'm not sure whether it's exactly correct. MRI may have short-circuited logic for !~ when one side or the other is a String or Regexp, and this would not reflect that short-circuiting.

diff --git a/src/org/jruby/parser/Ruby19Parser.y b/src/org/jruby/parser/Ruby19Parser.y
index 0830745..0562e8c 100644
--- a/src/org/jruby/parser/Ruby19Parser.y
+++ b/src/org/jruby/parser/Ruby19Parser.y
@@ -899,7 +899,7 @@ arg             : lhs '=' arg {
                   */
                 }
                 | arg tNMATCH arg {
-                    $$ = new NotNode(support.getPosition($1), support.getMatchNode($1, $3));
+                    $$ = support.getOperatorCallNode($1, "!~", $3, lexer.getPosition());
                 }
                 | tBANG arg {
                     $$ = support.getOperatorCallNode(support.getConditionNode($2), "!");
Show
Charles Oliver Nutter added a comment - Here's a hacky parser fix, but I'm not sure whether it's exactly correct. MRI may have short-circuited logic for !~ when one side or the other is a String or Regexp, and this would not reflect that short-circuiting. diff --git a/src/org/jruby/parser/Ruby19Parser.y b/src/org/jruby/parser/Ruby19Parser.y index 0830745..0562e8c 100644 --- a/src/org/jruby/parser/Ruby19Parser.y +++ b/src/org/jruby/parser/Ruby19Parser.y @@ -899,7 +899,7 @@ arg : lhs '=' arg { */ } | arg tNMATCH arg { - $$ = new NotNode(support.getPosition($1), support.getMatchNode($1, $3)); + $$ = support.getOperatorCallNode($1, "!~", $3, lexer.getPosition()); } | tBANG arg { $$ = support.getOperatorCallNode(support.getConditionNode($2), "!");
Hide
Permalink
Charles Oliver Nutter added a comment -

I incorporated a modified version of your spec, as well as sending a PR upstream to rubyspec/rubyspec.

Show
Charles Oliver Nutter added a comment - I incorporated a modified version of your spec, as well as sending a PR upstream to rubyspec/rubyspec.
Hide
Permalink
Hiro Asari added a comment -

Seems to me that Charlie's "hacky" fix is exactly what is happening in MRI: https://github.com/ruby/ruby/blob/7af3e9f08b6f277b7f96d7ed2a0115d1452d98cc/parse.y#L2340

We should commit that, I'd think.

Show
Hiro Asari added a comment - Seems to me that Charlie's "hacky" fix is exactly what is happening in MRI: https://github.com/ruby/ruby/blob/7af3e9f08b6f277b7f96d7ed2a0115d1452d98cc/parse.y#L2340 We should commit that, I'd think.
Hide
Permalink
Hiro Asari added a comment -

I pushed the fixed to the main branch: 5e78f58.

Show
Hiro Asari added a comment - I pushed the fixed to the main branch: 5e78f58.

People

  • Assignee:
    Charles Oliver Nutter
    Reporter:
    Jordan Sissel
Vote (0)
Watch (0)

Dates

  • Created:
    Updated:
    Resolved: