Closed lexaknyazev closed 1 year ago
This change seems reasonable, but I want to review it deeper before accepting.
This interacts with this report: https://github.com/BinomialLLC/basis_universal/issues/263
I'll need to set aside a day to make sure this doesn't break anything. I'll keep this PR up on my desk.
Any updates?
OK, this change should be good. The encoder should never emit tables with no symbols. Regression test works.
huffman_decoding_table::init
allowed code lengths like[0, 0]
, which should be invalid. Alternatively, if such inputs are valid this should be documented.