Hacker News new | past | comments | ask | show | jobs | submit login

The shared dictionary technique you describe is presently used for modern versions of HTTPS and for Brotli, IIRC. Very effective.



...though with the slightly unexpected side effect (for Brotli, at least) that your executable may end up containing (~200KB, from memory) of very unexpected[-1] plain text strings which might (& has[0]) lead to questions from software end-users asking why your software contains "random"[1] text (including potentially "culturally sensitive" words/phrases related to religion such as "Holy Roman Emperor", "Muslims", "dollars", "emacs"[2] or similar).

(I encountered this aspect while investigating potential size optimization opportunities for the Godot game engine's web/WASM builds--though presumably the Brotli dictionary compresses well if the transfer encoding is... Brotli. :D )

---- footnotes ----

[-1] An aspect also previously mentioned on HN: https://news.ycombinator.com/item?id=27160590

[0] "This needs to be reviewed immediately #876": https://github.com/google/brotli/issues/876

[1] Which, regardless of meaning, certainly bears similarities to the type of "unexpected weird text" commonly/normally associated with spam, malware, LLMs and other entities of ill repute.

[2] The final example may not actually be factual. :)


Just XOR it with 0x55 ;)


"Pros always use this one weird trick Brotli compression hates." :)




Join us for AI Startup School this June 16-17 in San Francisco!

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: