Weekly Update: Progress, Bug Fixes & The Road Ahead
Hi @everyone, apologies for the slowness of our weekly update today – just lots going on (and lots waiting in the wings). Let’s jump straight into it!
Release and Auto-Upgrade
As mentioned last time, a new release shipped last week and nodes have been auto-upgrading across The Network.
Performance: We tracked download speeds improving in real-time — dropping from 10 minutes down to 2.5 minutes for 15MB files.
Reliability: 100% download success rate once nodes reached the latest version.
Desktop App: v0.6.7 is live for Windows, macOS, and Ubuntu.
Fix: Resolved the Windows issue where nodes were incorrectly reporting as “stopped”.
Upload Bug: Found & Squashed
We identified a deterministic bug causing files of certain sizes to fail 100% of the time.
The Cause: A code mismatch in Merkle tree pool calculations for odd-depth trees.
The Fix: Now landed with 5 regression tests to prevent future occurrences.
Adaptive Client: Work is underway on a client that adjusts upload concurrency automatically. No more manual tuning—it should just work out of the box!
Client and Node Separation
The client and node codebases are now fully separated.
Why it matters: Updates for both can now be released independently. This allows for faster bug fixes and client-side improvements without requiring a wholesale node update.
Performance and Stability
Networking: Focused on routing table health and Kademlia iteration tuning (how nodes find each other).
Speed: Achieved a 3x download speed improvement by increasing the decryption batch size.
Validation: Currently running extensive multi-region testnets (900+ nodes) to ensure stability at scale.
What’s Next?
Our current focus is upload reliability for larger files. We are hunting down edge cases, including chunk rejections caused by misconfigured clocks on community nodes.
Once uploads are solid across all sizes, we unblock:
Reading Room content going live.
Indelible release.
Press & Partner work – everything is ready and waiting.
And with a small network at the moment so there will be chunks that are a long way from the downloader. And without any optimisations such as caching nodes which have been talked of for popular chunks. It is early days!
Please do no trouble @jane with realities such as yourself and @Warren just posted.
She has indigestion from having to eat humble pie and wants to moan.
This is a wonderful place to let her whine as due (in part) to her constant negativity, no bugger reads this forum anymore anyway.
Meanwhile the apps keep coming. @riddim just showed a couple of very interesting things, @feinberg feinbergis well advanced but saying little so far and there are plenty others.
We are now at the point where developers can really let rip on all these ideas that have bèn stewing for years. With AI to do the gruntwork of testing and GUIs, as Mao said, we will see a thousand flowers bloom.
Only the strongest will survive of course, but that’s nature and this project is built on nature.
This is a passage that suggests that Autonomi 2.0 could be not only a new form of the Internet but also a completely separate, decentralised network capable of operating without the involvement of operators, big tech companies, etc.
Is this same as the Reticulum network? Will have to open later.
Always positive to see these types of projects and their successes.
Feels like more projects are moving us closer to what Time Berners-Lee was trying to achieve with the SolidPod.
Which I’m all down for. If we build services that respect privacy, that “just work” and are reliable, people will choose them over the ones that do not.
V2.0 is looking promising. Looking forward to jumping back in.
These are the nodes for their test environment for the new versions, the live network has more nodes, from what people have shared on Discord I would say over 3000+
In fact, it doesn’t really matter whether there are 900 or 3,000 nodes; we are clearly at the stage of rolling out and testing the functionality of v. 2.0, and the most important thing is that the design principles hold up in practice. Only then will the time come for scaling and optimisation, and only then can we discuss network speed.