๐ We wrote the entire text of War and Peace into XRPL mainnet. All 3,339,794 bytes of it, split across 3,278 transactions, in 15.3 minutes.
Then we read it back out of the ledger and compared it to the original. Byte-for-byte match, SHA-256 verified.
๐ฐ IT COST FOUR CENTS
0.039336 XRP total. Every transaction paid 12 drops, and not one paid more.
XRPL's base fee does not scale with transaction size โ Transactor::calculateBaseFee charges the same for an empty transaction and one carrying a full kilobyte of memo. Bytes are free. For 3.19 MB that full-history nodes will now keep forever, the network charged about four cents.
That asymmetry is the whole experiment.
๐ THE PART WE DID NOT EXPECT
We rehearsed on testnet first. Mainnet was faster.
โธ Upload time: 20.84 min on testnet, 15.3 min on mainnet
โธ Ledgers consumed: 378 on testnet, 236 on mainnet
โธ Our share of a ledger, median: 90.9% on testnet, 8.3% on mainnet
On an empty testnet we were the traffic โ filling ledgers with ourselves, sometimes 100% of one. On mainnet the same 3,278 transactions dissolved into ordinary volume and rode along faster.
The network never noticed. open_ledger_fee stayed at 10 drops throughout, the transaction queue stayed empty, and ledgers held a median of 171 transactions against an expected size of around 733.
๐ WHAT THE PROTOCOL ALLOWS
โธ Memo ceiling is 1024 bytes for the whole serialized Memos array (STTx.cpp:695, isMemoOkay). Useful payload: 1019
โธ Adding an 11-byte MemoType tag pushes the transaction over the limit โ it fails local checks
โธ In the ledger each transaction occupies 1323.1 bytes: 1172.5 blob plus 150.6 metadata, or 1.298x the payload
โธ The queue accepts at most 10 transactions per account (TxQ.h:165)
โธ A default node keeps the last 256 ledgers (Config.h:209). "Forever" only applies to full-history nodes
๐ A COMPRESSION PARADOX
bzip2 squeezes the Russian text 5.72x, so it would fit in 909 transactions and 0.009 XRP. But compressed bytes look like noise to RocksDB, which cannot compress them again โ each stored byte then costs 1.5x more on disk. Net disk saving: 3.98x, not 5.72x.
Plain text is the opposite: RocksDB compresses it almost exactly as much as transaction overhead inflates it, landing at 1.048x.
๐ WE WERE PREDICTED
In XRPL-Standards discussion #103, someone wrote that memo examples would "land us in episode 143 of the 'I'll put my files on ledger in chunks' discussion" and that "sane people know this is a bad idea."
We are that episode. We just measured it.
This lands two weeks after the community argued over whether the memo limit should be raised at all โ one side holding that paying the fee should buy you the space, the other that every node stores it forever. Nobody put a number on it. Here is one: at today's limit, 1 GB works out to 1.05 million transactions and 12.64 XRP.
โ๏ธ Worth knowing: XRPL already has a per-byte fee, it is just not used for memos. An EscrowFinish carrying a fulfillment costs 330 drops plus 10 drops for every 16 bytes of the preimage. The mechanism exists. Memos simply do not use it.
๐ Read any chapter, assembled live from the ledger: https://t.co/wCeX2mIZwn
Code and dataset: https://t.co/Mdao9BRSMk
The site stores no copy of the book. Every character is fetched from transaction memos at read time and SHA-256 checked before display.
Account: rU1A1kuVpYHk3TZYWJxVtPUc94aXyteizy โ its entire history is the novel.
#XRPL #XRP #Tolstoy
We put xrpl-vanity on GitHub under Apache-2.0 ๐๏ธ
Itโs a vanity address generator for XRPL: you give it a word, it grinds keys on CPU or GPU until an address matches.
Runs locally, Ed25519 only.
Build it yourself first.
https://t.co/mlN6hks6fB
https://t.co/GSMGInc0Hw
@daniel_wwf@vjkhannaripple@Vet_X0@msvadari@XRPLF@bithomp@XamanWallet@XRP
FinishAfter: earliest time funds can go to the destination. CancelAfter: after it, they can only go back. Anyone can finish a ready escrow or cancel an expired one; funds only go where the escrow says.
A Condition needs the secret fulfillment: wrong one = tecCRYPTOCONDITION_ERROR. Finishing with it costs at least 330 drops + 10 per 16 bytes. Without CancelAfter, an escrow can never be cancelled.
Token escrow (since Feb 2026): the issuer must enable Allow Trust Line Locking, MPTs need Can Escrow. Always set CancelAfter if you may want the tokens back.
https://t.co/zaDMA5fkVQ
#XRP #Ripple #XRPL #Education
Carol finished an escrow that wasn't hers, and Bob got 10 XRP. That's how XRP Ledger escrow works.
3.5 minutes on FinishAfter, CancelAfter, crypto-conditions and token escrows, on a test ledger.
@XRPLF#XRPEducationMonth
@daniel_wwf@XamanWallet@LuckyHashWin@WietseWind@JoeyWallet no problem with WalletConnect.
Well, more precisely, of course, there are problems, but you need to understand what to do. We sent 4 PRs to their repository, conducted 300+ hours of testing to make it convenient for our users
@daniel_wwf@XamanWallet@LuckyHashWin@WietseWind@JoeyWallet I'm trying to understand why everyone limits the way they connect, Joey uses WalletConnect, but not only it, there are others in the ecosystem, but they don't know that they can connect.
Why not just specify WalletConnect?
3โฃ LP tokens are votes on the fee (0โ1%, weighted). Not guaranteed income: if prices move and stay, the loss can exceed the fees. https://t.co/30HKNkqTIe
1โฃ Deposit both assets in the pool's proportion and you get LP tokens, a share of the pool. On a test ledger: 100 + 100 gave 100,000 LP; a second provider's 50 + 50 gave 50,000. No fee for a two-asset deposit.
๐๐๐
Put 100 XRP + 100 USD into an XRP Ledger AMM, took out 101.18 XRP + 100 USD. No payment arrived. Where did it come from?
3 minutes on liquidity: LP tokens, when fees accrue, how to collect, and the risk. Real transactions on a test ledger.
@XRPLF#XRPEducationMonth
2โฃ Fees accrue inside every trade: they stay in the pool, nothing is paid out. 20 trades later the pool held 151.76 XRP for the same 150,000 LP. AMMWithdraw (tfWithdrawAll) returns your share, fees included.
๐๐
@daniel_wwf 12 wallets sent the same Batch into one XRP Ledger ledger. One got the trade, and the fee had nothing to do with it.
3.5 minutes on canonical order, open ledger escalation and why Batch swarms fill the cheap part of each ledger. On a test ledger.
@XRPLF#XRPEducationMonth
Loans with no collateral on the XRP Ledger: who takes the loss when one defaults?
5 minutes on the Lending Protocol (XLS-66, in voting): vaults and shares, first-loss cover, two-signature loans and the maths of a default, on a test ledger.
@XRPLF#XRPEducationMonth