Thanks for you talk!

I would like to ask one question and give two comments:
- Does it make sense to implement LZW with chaining? I thought the standard definition uses a trie. Of course, you can use hash fingerprints for faster navigation in the trie, but in the end, you want to find the longest match and therefore use the trie to navigate to a leaf. Can you elaborate on how your LZW approach works?
- It would be interesting to see an LZ77 factorization with an infinite window size (meaning without a window) to see how good the compression can actually become.
- Finally, about the "shorter hash chains", are you aware about the paper "Improving the Speed of LZ77 Compression by Hashing and Suffix Sorting" from Sadakane and Imai? (https://ci.nii.ac.jp/naid/110003208520/en)