The majority of last week I spent tinkering with Subspace codebase after importing it here in preparation for building an actual blockchain.
As mentioned in previous updates, transaction was the next logical step and after some preparation in PR 149 initial version landed in PR 151. It is quite basic and will likely require significant changes before used in an actual blockchain implementation, but there are just too many unknowns for now.
On a similar note, PR 156 added an initial version of the native token, which will be important for charging transaction fees. It is relatively simple for now, but in the spirit of what it will eventually end up being.
With that, Subspace codebase was imported in PR 152. This initial PR removed a bunch of features of the original codebase that will not be applicable here (domains, votes, etc.). There were many follow-up PRs that removed/refactored more things to reduce the feature set and reduce Substratisms from the codebase (PR 154, PR 155, PR 157, PR 158, PR 159). As the result the blocks can still be produced, but RPC is mostly gone (only supports what farmer needs) and many of the secondary features are gone too.
Blockchains are hard #
Now the challenge is figuring out where to go from there. Substrate is great because it gives a framework to work with, where an initial version of the blockchain may not be exactly what you want, but at least you have something working all the time. When bootstrapping from the ground up, it is easy to get lost since there are countless “chicken and egg” kinds of problems.
The plan right now is to first extract core consensus logic contained in the runtime into system contracts. After that to somehow get enough of node infrastructure up to get a single-node blockchain without P2P networking producing blocks.
P2P networking for blockchain will be a major effort since while Distributed Storage Network (DSN) in Subspace is standalone, block and transaction propagation, PoT gossips are all built on top of Substrate and were never migrated off of it. Given our sharding requirements, it is likely that this (originally) DSN networking stack will have to evolve to support more things, but that is a big challenge.
So yeah, blockchains are hard to pull of the ground, but when building something truly different, you sometimes have to. Thankfully, some lessons were learned, and the logic already written and audited in Subspace codebase can be partially reused to speed up the process (hopefully).
Upcoming plans #
It is hard to predict how such open-ended projects will go, but one way or another I’ll be working on building the blockchain to allow us experimenting with sharding that Alfonso is designing. One thing for certain is that you can expect regular progress updates along the way.
Until next time 👋