A Polkadot forum proposal swaps re-execution for proof checking
WonderZK would move heavy execution and proving off the validator set. It is one person's design, posted with a call for collaborators, not a project roadmap.
Everything we have published, newest first, from every desk.
WonderZK would move heavy execution and proving off the validator set. It is one person's design, posted with a call for collaborators, not a project roadmap.
A speech on market infrastructure names AI and tokenisation, puts trading hours in question, and commits to going as far as supervisory discretion allows.
BIP-174 gains a combiner vector for the sighash type, plus an instruction to run the existing vectors in both orders - because Combine is not commutative and the tests were.
Ethereum wants nodes to keep four months of beacon blocks instead of fourteen. Bitcoin wants to back up the data that is not the seed. XRPL is adding confidentiality to a token type. Retention is the common subject.
The upgrade went live at 11 AM ET on 22 September, bringing C-Chain async execution, auto-renewed staking, a higher uptime requirement, a shorter minimum stake, a dynamic gas floor and a lower consumption rate.
A new draft BIP defines a compact encryption scheme for output script descriptors, wallet policies and other metadata. The payload is forbidden from containing private key material.
Release No. IC-36333 grants ARK Venture Fund amended exemptive relief for two new share classes, one of which can record ownership on distributed ledger. Nothing is trading yet.
The autumn 2026 joint risk update from EBA, EIOPA and ESMA puts external dependencies, private credit and AI-enabled cyber threats at the top, while reporting the EU financial system as resilient.
The draft lowers MIN_EPOCHS_FOR_BLOCK_REQUESTS from 33,024 epochs to 8,192, arguing the existing figure comes from a worst-case weak-subjectivity calculation nobody actually uses.
The leader-info syscall proposal now documents pointer validation step by step: compute units first, then an unaligned-loader guard, then a direct rejection of input-region destinations.