Skip to content

token_stream and skip_document inconsistency #1492

Open

Description

@tompng

Complex#conjugate have token_stream but marked as skip_document

Complex#conjugate has ▶︎Source toggle but description is skipped.
Complex#conj has document but ▶︎Source is not available.

The behavior of this bug looks similar to #1335 but the cause is different.

skip_description

# Whether to skip the method description, true for methods that have
# aliases with a call-seq that doesn't include the method name.

I think this design decision makes sense. There is a room for discussion whether we really need to generate duplicated document for the same method.

token_stream

Among method/aliases with the same source, RDoc seems to select one main method. token_stream is associated to that method object.
RDoc shouldn't select description-skipped method object as main method.

Reproduction code

/*
 * call-seq:
 *   conj -> complex
 * Description
 */
VALUE rb_complex_conjugate(VALUE self){return Qnil;}
void Init_Foo(){
    rb_cComplex = rb_define_class("Complex", rb_cObject);
    rb_define_method(rb_cComplex, "conjugate", rb_complex_conjugate, 0);
    rb_define_method(rb_cComplex, "conj", rb_complex_conjugate, 0);
}

RDoc thinks the first define_method (conjugate) is the main definition. token_stream is associated to it. But conjugate is not in call-seq.

Methods that have this problem

Complex#conjugate
Complex#imaginary
Complex#rectangular
Enumerable#collect
Enumerable#find_all
Enumerable#member?
Forwardable#instance_delegate
Hash#initialize_copy
IO#each
Integer#===
MatchData#eql?
Numeric#conjugate
Numeric#imaginary
Numeric#rectangular
ObjectSpace::WeakMap#include?
Range#member?
Regexp#eql?
SingleForwardable#single_delegate
StringIO#each
StringIO#eof
Time#getgm
Time#gmtime
Time#gmtoff
Time#isdst
Time#tv_nsec
Time#tv_usec

Activity

  1. extern-c commented on Feb 28, 2026

    @extern-c
    Contributor

    Hi @tompng. I started looking into this and wanted to confirm: is this considered a bug, or is this the intended behavior in RDoc? If it's the latter, would updating the call-seqs for those methods to match the current behavior be the preferred approach? If this isn't already documented, we could also clarify it in the documentation. This does appear to be a bug.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Type

    No type

    Projects

    No projects

      Milestone

      No milestone

      Relationships

      None yet

      Development

      No branches or pull requests

      Issue actions