Research Substrate's pallet-scheduler

[razgraf]

While researching on the topic of proactive blockchains I've found this interesting utility part of the Substrate/Polkadot stack.

Substrate is a blockchain building framework which serves as the core architecture for Polkadot. To extend its core functionality one can start from FRAME (the Framework for Runtime Aggregation of Modularized Entities) which enables customization and includes the topic of this conversation: the pallet-scheduler.

From its description:

This module exposes capabilities for scheduling dispatches to occur at a specified block number or at a specified period. These scheduled dispatches may be named or anonymous and may be canceled.

This could be useful (as a tool or simply as a mental model) for our pre-scheduled insolvency deadlines.

Video explainer

https://www.youtube.com/watch?v=olcuZHb4u88

At 10:12 there's an interesting note around the MaximumScheduledPerBlock parameter. This feels relevant to us as well: what happens if there are a lot of deadlines ending in the same block? How could they all be settled (on the spot) or can we find a lazy way of settling them regardless of the post-deadline block they're included in.

ChatGPT explainer (what prompted this research)

Note the use-cases at the end.

CC: @sablier-labs/sabvm

[PaulRBerg]

Interesting finding, and thanks for sharing, though I don't see how this is directly relevant to us.

There might be some value in peeking at their implementation (which is also Rust-based) to get some inspiration on how to schedule cron jobs in a blockchain environment, but generally speaking, this Pallet Scheduler thingy isn't by itself a solution to the Concentrated Termini DoS problem. Iaroslav and I dived deep into the problem in the linked discussion, and we already have a battle plan for how to tackle it.

---

While researching on the topic of proactive blockchains

It does seem as if we weren't the first. I've also heard of some gaming rollups needing to perform event-driven computations.

Nonetheless - we can be the first to own the "Proactive Blockchains" meme and turn it into a defining attribute of Sablier.

Celestia wasn't the first DA layer, but they captured the "Modular Blockchains" and "DA" memes.

---

the pallet-scheduler.

This link redirects to https://substrate.io/ now

---

note around the MaximumScheduledPerBlock parameter. This feels relevant to us as well: what happens if there are a lot of deadlines ending in the same block?

Yeah, it is relatable, but doesn't offer any solution.

Their tasks are meant to be defined by users and are thus arbitrarily sized. This makes me speculate that they went for a low value for MaximumScheduledPerBlock.

In our case, setting a user's balance to zero is a deterministic action, which is why we can benchmark it.

[PaulRBerg]

I will let you close this discussion @razgraf (if you agree with my assessment).

[razgraf]

Sure we can close it, this was part of my initial research that led to the observations in #90. Will continue discussing there and reference anything from the Polkadot repository if it fits our needs.