How do you tell a chain is close to mainnet when the announcement hasn't landed? Not from the roadmap. From what the engineers touched last week.
Start with my own week. I deployed a plain Solidity contract to DuskEVM testnet, chain 745, with Foundry. forge create against the testnet RPC, then opened it in the explorer. Nothing to relearn. That's the whole pitch of the EVM lane, and it holds up.
Then I read @Dusk 's developer update for August 17 to 24. Eleven public repos. Almost no features.
PLONK took the bulk of it. Bounded compressed-circuit allocation, rejected malformed indices, polynomial lengths, public-input positions and curve-addition edge cases. Cached prover and verifier data, grouped multiscalar work, parallelised independent FFT stages. Every entry ends the same way: proofs, transcripts and serialization outputs unchanged.
Poseidon252 now rejects invalid encryption shared secrets before key derivation, covering identity, off-curve, low-order and mixed-order points.
The Web Wallet shipped three things a user actually feels. A 15-minute inactivity lock that enforces its deadline even in a background tab. Versioned mnemonic encryption with a raised PBKDF2 work factor. Migration for legacy records.
And duskevm-genesis extended a disposable devnet rehearsal's sequencing recovery window by 400 L1 blocks, without resetting the rehearsal history.
Rehearsal. That's the word I'd underline.
My take: it reads like a team freezing formats and practising recovery before anyone settles securities on top. Boring commits are the tell.
The less flattering read of the identical list: dependency bumps and test refactors also describe a team with nothing shippable. From the outside I can't fully separate the two. Chain 744 sits in the docs, the launch post still doesn't. $DUSK pays gas on both lanes and secures the L1 through staking.
So, back to where I started. When you're sizing up a chain that hasn't launched yet, what do you read first?
#dusk
Start with my own week. I deployed a plain Solidity contract to DuskEVM testnet, chain 745, with Foundry. forge create against the testnet RPC, then opened it in the explorer. Nothing to relearn. That's the whole pitch of the EVM lane, and it holds up.
Then I read @Dusk 's developer update for August 17 to 24. Eleven public repos. Almost no features.
PLONK took the bulk of it. Bounded compressed-circuit allocation, rejected malformed indices, polynomial lengths, public-input positions and curve-addition edge cases. Cached prover and verifier data, grouped multiscalar work, parallelised independent FFT stages. Every entry ends the same way: proofs, transcripts and serialization outputs unchanged.
Poseidon252 now rejects invalid encryption shared secrets before key derivation, covering identity, off-curve, low-order and mixed-order points.
The Web Wallet shipped three things a user actually feels. A 15-minute inactivity lock that enforces its deadline even in a background tab. Versioned mnemonic encryption with a raised PBKDF2 work factor. Migration for legacy records.
And duskevm-genesis extended a disposable devnet rehearsal's sequencing recovery window by 400 L1 blocks, without resetting the rehearsal history.
Rehearsal. That's the word I'd underline.
My take: it reads like a team freezing formats and practising recovery before anyone settles securities on top. Boring commits are the tell.
The less flattering read of the identical list: dependency bumps and test refactors also describe a team with nothing shippable. From the outside I can't fully separate the two. Chain 744 sits in the docs, the launch post still doesn't. $DUSK pays gas on both lanes and secures the L1 through staking.
So, back to where I started. When you're sizing up a chain that hasn't launched yet, what do you read first?
#dusk
