Misc #20994: C API - How to Create Classes with Non-ASCII characters in Name

• Status: Closed
• Assignee: -

[cfis (Charlie Savage)]

As part of Rice (Ruby C++ bindings), I am experimenting with using unicode characters to make more readable class names (see https://ruby-rice.github.io/4.x/stl/stl.html#automatically-generated-ruby-classes).

I am experimenting with class names like this:

Map≺string≺char≻٬vector≺complex≺double≻≻≻`

Where < and > are actually Unicode characters precede (\u227A) and succeed (\u227B).

In Ruby this works fine:

irb(main):01> class Map≺string≺char≻٬vector≺complex≺double≻≻≻
irb(main):02> end
=> nil
irb(main):013> Map≺string≺char≻٬vector≺complex≺double≻≻≻.new
=> #<Map≺string≺char≻٬vector≺complex≺double≻≻≻:0x0000021114674c98>

However, this fails using the Ruby C API rb_define_class. Passing a char* that is utf8 encoded fails because rb_define_class calls rb_intern which calls rb_intern2 which forces the use of ascii encoding (see https://github.com/ruby/ruby/blob/5fec9308320e8b377681ef19b0cd46d53f94e8ac/symbol.c#L818).

I thought I might be able to define the class using ascii characters, and then call rb_define_const to add in a utf8 encoded name, but that also has the same problem.

My question - how does one create classes that have non ascii characters in their names them via the C API?

[cfis (Charlie Savage)]

• Description updated (diff)

[cfis (Charlie Savage)]

• Description updated (diff)

[cfis (Charlie Savage)]

Ah - I found rb_define_class_id_under which works. This ticket can be closed.

[nobu (Nobuyoshi Nakada)]

• Tracker changed from Bug to Misc
• Status changed from Open to Closed
• ruby -v deleted (ruby 3.3.6 (2024-11-05 revision 75015d4c1f) [x64-mswin64_140])
• Backport deleted (3.1: UNKNOWN, 3.2: UNKNOWN, 3.3: UNKNOWN, 3.4: UNKNOWN)

Or you can assign an anonymous class to that constant.