Open laughinghan opened 4 years ago
I did a test with this for nbvcxz (Java port) just for reference, and it made virtually no difference in compressed size of the default dictionaries. It may be more useful for this version, but it wasn't worth the effort when I looked into it for my project.
It would also probably save some CPU cycles to not need to parse the giant string into an array on load
The wordlists are huge. Gzip and async loading helps of course, but succinct data structures are made specifically for this purpose and should work better than gzip: http://stevehanov.ca/blog/?id=120