1.7.0 Release
This release was started as performance improvement. But since work on that improvement took a while other updates and features came in as well.
Change set:
- Database engine switched to RocksDB.
Prefix lookups feature supported by RocksDB allowed us to get rid of reference counting and make database flushes 100 times faster improving speed of blocks processing on the Mainnet. - Pruning completely re-worked.
Previous algorithm was based on reference counting and we had to build something different. New algorithm uses bloom filters and gives 99% prune rate with almost zero processing time. - Snappy compression support was added.
Check EIP-706 for details. - Project moved to Java 8.
- Implementation of Event API was done (#421)
New API provides easy and efficient access to event and log management. - An option to skip bodies and receipts was added.
Fast sync just finishes much faster without fetching all this big data. - Memory consumption of fast sync queues was limited.
Now fast sync is able to get finished with-Xmx1g
on the Mainnet. - Several bug fixes were made.
- NOTE: db engine has been changed, hence an update to this version requires database reset.
Thanks to community for their help: