Fix Encoding::ConverterNotFoundError

koic · koic · commit 6df6e76cfc9f · 2021-05-11T11:31:53.000+09:00
Follow #237. This PR fixes the following `Encoding::ConverterNotFoundError`. ```console % bin/spring stop && bin/rails c Spring stopped. Running via Spring preloader in process 58395 Loading development environment (Rails 6.0.3.7) irb(main):001:0> "こんにちは".do_something Traceback (most recent call last): (snip) 12: from /Users/koic/src/github.com/ruby/irb/lib/irb.rb:547:in `eval_input' 11: from /Users/koic/src/github.com/ruby/irb/lib/irb/ruby-lex.rb:232:in `each_top_level_statement' 10: from /Users/koic/src/github.com/ruby/irb/lib/irb/ruby-lex.rb:232:in `catch' 9: from /Users/koic/src/github.com/ruby/irb/lib/irb/ruby-lex.rb:233:in `block in each_top_level_statement' 8: from /Users/koic/src/github.com/ruby/irb/lib/irb/ruby-lex.rb:233:in `loop' 7: from /Users/koic/src/github.com/ruby/irb/lib/irb/ruby-lex.rb:251:in `block (2 levels) in each_top_level_statement' 6: from /Users/koic/src/github.com/ruby/irb/lib/irb.rb:548:in `block in eval_input' 5: from /Users/koic/src/github.com/ruby/irb/lib/irb.rb:758:in `signal_status' 4: from /Users/koic/src/github.com/ruby/irb/lib/irb.rb:586:in `block (2 levels) in eval_input' 3: from /Users/koic/src/github.com/ruby/irb/lib/irb.rb:650:in `handle_exception' 2: from /Users/koic/src/github.com/ruby/irb/lib/irb.rb:601:in `encode_with_invalid_byte_sequence' 1: from /Users/koic/src/github.com/ruby/irb/lib/irb.rb:601:in `new' /Users/koic/src/github.com/ruby/irb/lib/irb.rb:601:in `initialize': code converter not found (UTF-8 to UTF-8) (Encoding::ConverterNotFoundError) ``` First, this patch skips `Encoding::Converter.new` for the same encoding. https://github.com/ruby/irb/blob/170531df19bce289444afe97360480efed5f27f0/lib/irb.rb#L601 Next, this is a talk about the condition for skipping. `IRB.conf[:LC_MESSAGES].encoding` becomes `"UTF-8"` string when `Reline.encoding_system_needs.name` is set in the below. https://github.com/ruby/irb/blob/170531df19bce289444afe97360480efed5f27f0/lib/irb/input-method.rb#L269 OTOH, `message.encoding` is `Encoding::UTF_8`, so these are compared as a string by this patch.
diff --git a/lib/irb.rb b/lib/irb.rb
@@ -647,7 +647,7 @@ def handle_exception(exc)
           order = :top
         end
         message = convert_invalid_byte_sequence(message, exc.message.encoding)
-        message = encode_with_invalid_byte_sequence(message, IRB.conf[:LC_MESSAGES].encoding) if message.encoding != IRB.conf[:LC_MESSAGES].encoding
+        message = encode_with_invalid_byte_sequence(message, IRB.conf[:LC_MESSAGES].encoding) unless message.encoding.to_s.casecmp?(IRB.conf[:LC_MESSAGES].encoding.to_s)
         message = message.gsub(/((?:^\t.+$\n)+)/)  { |m|
           case order
           when :top
