Dominic Cooney

Happy to review... could you look at the failing test first?

The graphs on p. 10 and 11 seem to indicate that TreeRePair produces smaller output.

Not sure if this is useful, but I've done some analysis of how effective Brotli is on top of TreeRePair. Some knowledge of Brotli is assumed for brevity but please...

I'm starting to do some similar analyses on these now. I was imagining something different for "MRU", more like the explicit labeler but reserving 0..n (probably with pretty small n)...

I played with trp-rank and there's exponentially diminishing returns and compression time increases. I see the counter productiveness of MRU but I haven't dug into explaining it. I tried doing...

OK, I've sketched some ideas [in the commits here.](https://github.com/dominiccooney/binjs-ref/commit/058978db1ab6a0583d8262402d2a1ad386bc97b9) I'm new to Rust and have no idea what I'm doing ;) Using `compare_compression -c identity --dictionary header --format multistream` on...

Brief update, I've tried implementing Burrows-Wheeler [(commit)](https://github.com/dominiccooney/binjs-ref/commit/0bdeeb7853f9d9d245f7bbf9a7ab9eadec3bc022) but it seems counterproductive for Brotli. With the sorted string table + BWT the size ratio regresses from 1.142 → 1.209. Guess there's...

I did some model development, based on these assumptions: - The encoder performance is a linear combination of a "dictionary hit" compression ratio and a "dictionary miss" compression ratio. I've...

Sorry for the slow reply, I missed this earlier. > > Focus on producing better compression than Brotli when hitting a dictionary of any size. > > I have difficulties...

Per [_ExpressionStatement_](https://www.ecma-international.org/ecma-262/#sec-expression-statement) grammar, that is parsed as a block, not an expression statement, hence : is a syntax error. It looks to me that the original file is actually JSON...