Open clstrfsck opened 5 years ago
@mpsandiford Just to make sure I understand this correctly: I think you are suggesting addition of a CBORGenerator.Feature
that would force use of length-prefix for all Strings. Is that correct?
That sounds reasonable, I just want to make sure I understand this correctly.
@cowtowncoder — I hadn't thought of it that way, but yes, that sounds like a great way to do it. That way there is an option for "always length prefix" or the more performant "indefinite" encoding.
Ok cool. As I said, it may take a while to get this done if I do it (would be in 2.10.0 regardless, being new feature) just due to piling up of issues. But I can help with a PR if that is something you could help with? (I try to prioritize helping others above my own work)
Yep, sure, sounds like a good approach. Give me a couple of days and I'll submit a PR with the feature and we can take it from there.
In
CBORGenerator.java
, there is a limit to the length of the string that can be output before the generator will use the indefinite length encoding. The reason behind this appears to be the complexity of determining the length of the encoded byte stream when it does not fit into an internal buffer, given that the length prefix needs to be output before the data.For now, we have changed the implementation of
_writeChunkedString(...)
to the following:It should also be possible to use the existing UTF-8 encoder to write the encoding out in pieces with the addition of a
int encodedLength(...)
method to precompute and output the length.Also, thanks for the awesome library!