There appears to be a regression in transcoding UTF-8 -> Windows-1255, introduced sometime after jcodings 1.0.26. The following Ruby code (taken from the MRI test suite), demonstrates the problem:
"\xCA".encode("utf-8", "windows-1255")
With jcodings 1.0.26, this will result in an Encoding::UndefinedConversionError being raised. With jcodings 1.0.28, no error is raised and an empty string is produced instead.
There appears to be a regression in transcoding UTF-8 -> Windows-1255, introduced sometime after jcodings 1.0.26. The following Ruby code (taken from the MRI test suite), demonstrates the problem:
With jcodings 1.0.26, this will result in an
Encoding::UndefinedConversionError
being raised. With jcodings 1.0.28, no error is raised and an empty string is produced instead.