I measured what a Finalis node actually writes to disk on testnet.
The chain was the small part. On one host the logs alone had grown to 2.5 GB — five times the size of the chain.
A merge request still in review cuts log lines by ~90% and caps RocksDB’s own logs.
#blockchain #nodes
Next: every transaction is stored three times in the database. Removing the extra copies should cut chain data by roughly half (my estimate from today’s breakdown — I’ll confirm with a real measurement and post the number).
Small nodes on cheap disks matter if ordinary people are going to run validators.
5
You don't need to write code to help Finalis. Bug reports ("I didn't get this screen" counts), testing on cheap Android phones, translating the site, tutorials.
Where to start:
gitlab.com/finaliscore/final…
1
2
24
Rules for money, written before there is any money: paid to whoever did the work, never to me or through me. One public ledger, approvers who aren't me, anyone can object. No funding channel yet.
gitlab.com/finaliscore/final…
11
George contributed to the Finalis website and then asked whether we’d open a fully transparent sponsorship channel (Open Collective / Liberapay / etc.) so early contributors and non-coders can support open-source work before mainnet rewards begin—never as an investment, every payment public, consistent with the no-premine rules.
This is how building in public actually works: one honest commit, one careful question, gradual trust. The code, the constraints, and the conversation are all visible.
gitlab.com/finaliscore/final…
#OpenSource #BuildInPublic #GitLab #Crypto
finalis retweeted
Public key cryptography might stand on very narrow foundations soon (especially the post-quantum kind)
Home nodes just got a lot more usable on Finalis.
v2.3.5 is out on testnet-3. A node can now ask the router (via PCP) to open its port, and open an IPv6 pinhole when your ISP puts you behind CGNAT.
One more reachable home node = one less dependency on my servers.
gitlab.com/FinalisCore/final…
1
2
34
Ran it on my own home router. It correctly called out no PCP, UPnP refused, and no IPv6 route, then told me what that router actually needs.
If you’re on CGNAT with IPv6, reply with your output. I want to see how well it works in the wild
1
8
Finalis Core v2.3.4 is out on testnet-3. Any node can now serve wallets on its P2P port, encrypted to the node's own key: no domain, no certificate, one forwarded port.
gitlab.com/FinalisCore/final…
How it works: the node announces its address signed with its key. The wallet connects with Noise NK to that key, so only that node can answer. No certificate authority is involved.
1
19
Huge thanks to @gitlab for welcoming our project into the GitLab for Open Source program! 🚀 Excited to leverage GitLab Ultimate's DevSecOps, security scanning, and CI/CD tools to build open, community-driven software.
Check out our repo: [gitlab.com/finaliscore]
finaliscore.org
#OpenSource #GitLab #DevSecOps
Finalis Wallet v0.0.23 for Android is out.
The wallet no longer just trusts the server. It checks every coin against validator signatures, starting from genesis.
It asks up to 5 servers. One honest one is enough.
Testnet only.
gitlab.com/finaliscore/final…
Finalis Core v2.3.2 is out on testnet-3. Home nodes behind a router got most of the work: IPv6 peers, verified announced addresses, and a status that tells you if others can reach you.
gitlab.com/FinalisCore/final…
1
49
Checks: the whole testnet-3 chain replayed from block 0 with v2.3.2 reaches the same blocks. A live test on v2.2.16: 221 transactions of 10 types, all finalized, none rejected. clang-tidy findings: 158 to 0.
1
11
Running a node at home: you don't need to open anything, but every reachable home node makes the network harder to cut off. Guide for routers, carrier-grade NAT and IPv6:
gitlab.com/FinalisCore/final…




