Testing and monitoring continues apace and the good news is the network seems more stable. Uploads and downloads are working, and we can’t find any data loss. App developers are reporting a better experience overall although there are still some questions around scratch pads and pointers that we’re looking into. Not quite sure whether there is a fault or not, but we are looking into it.
In terms of network stability, we’re still working on reachability checks to make sure that nodes are actually reachable, and to remove nodes that are not reachable by the rest of the network. We are also removing the relay nodes because these are dragging down other nodes as they use operator’s bandwidth. Initial results are positive.
As mentioned last week, we ran an internal hackathon which gave the developers a chance to do a bit of dogfooding and work on the API. Safe to say, we learned a LOT.
The self_encryption library has been refactored and now supports streaming so we can handle much larger files, or any size of file really. Again, that’s looking promising and we should soon be ready to add it into the client libraries.
Our most recent release was a client-only hotfix which restored uploads and downloads to working order and added support for resuming downloads. So if you have a download where one chunk fails, you can just run it again and you only need to download the single chunk.
Community happenings
@happybeing released the latest dweb (v0.10.8) which he says will improve Friends, Atlas and other web apps using the dweb REST API.
@zettawatt released Colony v1.1.2 with several UI improvements and fixes. macOS binaries are now properly signed and wi@riddiml run on any Mac computer.
@riddim put out a new release of Friends with encryption right from the first friendship
And @traktion released a milestone v0.10.0 of AntTP including Tarchive (tar archive) integration and improved upload performance.
General progress
Alphabetical this week.
@anselme has been working on streaming project and started integration into autonomi so we can quickly bring streaming (support for big files) to the community and developers.
@bzee has been working on a connectivity check based on ConnectionEstablished events received to ensure our client is properly established.
@chriso changed the self_encryption release process to a semi-automated setup that will have human review. He also simulated static full cone NAT hosts in testnet-deploy, based on shu’s research. This means we can deploy testnets with a greater proportion of port-forwarded nodes which will be useful for staging comparisons and the new alphanet that will be used for testing. We can also simulate UPnP hosts in testnet-deploy to improve our QA. In addition Chris set up a continuous downloader for the production network that goes through our metrics system to enable us to collect data on download performance.
Ermine investigated a download issue in the hackathon client when connected to the mainnet, and also worked on enabling port-restricted cone NAT for nodes in the testnet.
Lajos has been working on the voters rewards system for Impossible Futures.
@mick.vandijke has been looking into how we might be able to reduce gas fees. He found an issue with silently failing file uploads and raised a PR to fix it. Mick also created a PR to store data addresses and private datamaps in the local user_data folder after uploading like@rolandwe do for@rolandmulti-file uploads (archives).
@roland raised a PR to expose antnode_branch as a metadata field, and modified the code to get the listen_addr from p2p before starting the reachability check process. He reports that things are working great. Roland also continued removing RPC from antnode manager + launchpad integration.
@qi_ma has been on streaming and self_encryption. He raised a PR to make tests for backward compatibility of the self_encryption changes.
@shu made some commits for the testnet_report tool to support the nat_static_full_cone_node host role.
