Conversation

@cwebber so you still have to read char by char, constructing a number until you see a ':' then you read that many bytes as a data element? That seems oddly redundant - if you're scanning for delimiters anyway why do you . need the counts?
I've seen offset+tag counted files like IFF (WAV, AIFF, mov etc) when they're binary, and COBS that does it with octets http://www.stuartcheshire.org/papers/COBSforToN/
I'm not sure what the length + colon is providing compared to TSV or JSON