Release v0.16.6-rc.4 · NethermindEth/juno

This release brings support for Starknet 0.14.4 and includes breaking changes that make it incompatible with Starknet 0.14.3.

We've refreshed and updated our docs, and Streaming is back on the menu as the default way to download a Juno snapshot.

Additionally, many performance and security improvements have been implemented.

Added

Pre-confirmed block RPC requests have access to declared classes as well. #3821
Awareness to Juno about how much memory it has available for calculating compilation defaults. #3764
RPC timeout now applies to both http and websocket endpoints (before, it was only http). #3870
Internal architecture update for flexible L1 clients (not fixed to ETH). #3738
New flag to start the node with synchronization disabled. #3912
Pool of compression writers (expensive allocation) to reuse writers when giving compressed responses. #3956
- Also, added a faster Gzip engine #3957
Updated to use Go 1.27.

Increased --db-max-handles default values to half of the fd limit given to the process (previously fixed at 1024). This increases Juno database access speed. #3919
Optimized database access for many methods and accessors to just load the important bits (literally!). #3810 #3809 #3818 #3833 #3834 #3835 #3857 #3876 #3880 #3886 #3885 #3890 #3888 #3895 #3889 #3916 #3918 #3904 #3909 #3917 #3961 #3952
Change many reference types to value types (in types definition and function signatures), lowering Juno's natural heap allocation. #3843 #3858 #3906 #3938 #3908 #3955
CBOR encodings and decodings for many internal types. #3792 #3796 #3850 #3951 #3953
JSON encoding/decoding for felts. #3831 #3881

Juno now answers correctly CONTRACT_NOT_FOUND when querying for system contracts, matching the spec. #3788
Prune batches (when the node is running with --prune-mode are properly closed on error. #3905
Sync status is now reported correctly when the node is run with --prune-node. #3864
Bound Gzip decoding to avoid a decompression bomb. #3976

Full Changelog: v0.16.5...v0.16.6-rc.4