This is a follow up issue to discussions in #455 in regards to how encodings should be handled. In that PR, I tackled the correction of the value transcoding based on header character set. It was identified that the current character set detection has the potential to corrupt values if multiple character sets are requested (first at the global header and then within an SR node).
The correct handling would to use "partial specialization subclasses that detect a particular variant and convert the dataset into a form that you would expect according to the standard." Using this pattern, we could temporarily switch the transcoding encoding to handle special cases like SR nodes.
Note: There are many SR objects in the wild switching character sets within nodes.
This is a follow up issue to discussions in #455 in regards to how encodings should be handled. In that PR, I tackled the correction of the value transcoding based on header character set. It was identified that the current character set detection has the potential to corrupt values if multiple character sets are requested (first at the global header and then within an SR node).
The correct handling would to use "partial specialization subclasses that detect a particular variant and convert the dataset into a form that you would expect according to the standard." Using this pattern, we could temporarily switch the transcoding encoding to handle special cases like SR nodes.
Note: There are many SR objects in the wild switching character sets within nodes.