A Java parser/writer for the ETF format.
This api aims to provide a set of high-speed abstractions for ETF. Additionally, it supports multiple subsets of ETF: standard ETF, BERT-rpc and Hammer & Chisel's (Discord's) Loqui.
TODO
While this project attempts to be as well optimized as possible, the JVM creates a limit to how fast this can be just due to how it manages memory.
Performance can be improved via JNI or usage of the Unsafe class
but these have issues which has kept me using pure java.
Regarding JNI: I do not know enough C/C++ to properly optimize and there would still be a cost for interfacing natives with java (meaning that for smaller ETF data there won't be much of a benefit).
Regarding Unsafe: Unsafe is unsafe. Meaning it is not guaranteed to
work across java versions and jvm implementations. So despite its
significant performance benefits, it's impossible to be sure that it will
always work how I expect it to.
Due to java not implementing proper primitive unsigned data types, I have
been forced to use much larger data types than should be necessary to
prevent potential underflows (ex. long instead of int or BigInteger
instead of long). This causes some strange internal data type handling
and some slight overhead.