Bug #21029: Prism behavior for `defined? (;x)` differs

• Status: Assigned
• Assignee: prism
• Target version: -
• ruby -v: ruby 3.5.0dev (2025-01-11T03:21:57Z master 1b3037081e) +PRISM [x86_64-linux]
• Backport: 3.1: UNKNOWN, 3.2: UNKNOWN, 3.3: UNKNOWN, 3.4: UNKNOWN

[qnighy (Masaki Hara)]

Prism has a different behavior for (;expr) when used in defined? predicate:

% ./miniruby --parser=prism -e "p defined? (;x)"
nil
% ./miniruby --parser=parse.y -e "p defined? (;x)"
"expression"

Although not a significant difference, aligning either of them with the other would be better.

[kddnewton (Kevin Newton)]

• Assignee set to prism

[kddnewton (Kevin Newton)]

I think we should change parse.y to match this behavior. I will ask around.

[hsbt (Hiroshi SHIBATA)]

• Status changed from Open to Assigned

[matz (Yukihiro Matsumoto)]

I think compound expressions (expressions (including empty) concatenated by semicolons) should be “expression” as parse.y.
It makes defined? simpler. I know the current defined? behavior recursively check for defined-ness (e.g., method parameters), I don't think we need that complexity here.

Matz.

[qnighy (Masaki Hara)]

I think compound expressions (expressions (including empty) concatenated by semicolons) should be “expression” as parse.y.

Interestingly enough though:

% ruby --parser=prism -e "p defined? (x;)"
nil
% ruby --parser=parse.y -e "p defined? (x;)"
nil

% ruby --parser=prism -e "p defined? (;x)"
nil
% ruby --parser=parse.y -e "p defined? (;x)"
"expression"

[kddnewton (Kevin Newton)]

@matz (Yukihiro Matsumoto) for what it's worth, it makes it much more complicated in the Prism compiler because we don't have empty statements. So it's not a recursive situation at the moment, it's just a single statement.

As @qnighy points out, should this be "expression" for when there is a trailing ;?