Is DecodingError for byte \0x80 expected?

[grzuy]

$ irb
irb(main):002:0> require "libcbor/all"
.../.gem/ruby/2.6.0/gems/libcbor-0.1.0/lib/libcbor.rb:50: warning: constant ::Fixnum is deprecated
=> true
irb(main):004:0> CBOR.decode(CBOR.encode("\x7F".b))
=> "\x7F"
irb(main):005:0> CBOR.decode(CBOR.encode("\x80".b))
Traceback (most recent call last):
        7: from ../ruby-2.6.0/bin/irb:23:in `<main>'
        6: from ../ruby-2.6.0/bin/irb:23:in `load'
        5: from ../ruby-2.6.0/lib/ruby/gems/2.6.0/gems/irb-1.0.0/exe/irb:11:in `<top (required)>'
        4: from (irb):5
        3: from ../.gem/ruby/2.6.0/gems/libcbor-0.1.0/lib/libcbor.rb:65:in `decode'
        2: from ../.gem/ruby/2.6.0/gems/libcbor-0.1.0/lib/libcbor.rb:65:in `tap'
        1: from ../.gem/ruby/2.6.0/gems/libcbor-0.1.0/lib/libcbor.rb:65:in `block in decode'
CBOR::DecodingError (CBOR::DecodingError)

[cabo]

I can't build libcbor right now. What does CBOR.encode("\x80".b) give you?

[grzuy]

> cbor = CBOR.encode("\x80".b)
=> "a\x80"
> cbor.encoding
=> #<Encoding:ASCII-8BIT>

[cabo]

Oh. That means the encoder does not recognize "\x80".b as a byte string and instead encodes it as a text string. "\x80" is not valid UTF-8, so this might be why the decoding then fails.

[cabo]

Apparently, libcbor-ruby does not use the encoding to distinguish byte from text strings, but requires the user to use the separate CBOR::ByteString class for byte strings. I can't test this right now, but if you replace "\x80".b by CBOR::ByteString.new("\x80".b), this should work.

[grzuy]

Indeed, using CBOR::ByteString.new("\x80".b) works.
Thank you!

Would be great if .encode could be able to distinguish byte from test string...

[cabo]

You could modify CBOR::StringHelper::__libcbor_to_cbor (lib/helpers.rb) to check for encoding and do what CBOR::BytestringHelper::__libcbor_to_cbor does if the encoding is BINARY. (If legacy encodings are still important for text strings for you, you could even convert to UTF-8 for those encodings.)

[grzuy]

Thank you

[PJK]

Yup, @cabo is spot on. Filed #6 to improve this.