Member Request

[koute]

Hi.

I'm Jan, and I have been directly contributing to Polkadot (counting from my first PR) since July 2021.

My main contributions are in the area of performance, and anything and everything WASM-related, but I also try to dabble in other areas too from time to time.

First PR

paritytech/polkadot#3519 (a memory leak fix in Polkadot)

Notable PRs

A lot of PRs improving performance:

• Speed up timestamp generation when logging paritytech/substrate#9933
• Speed up logging once again paritytech/substrate#9981
• Remove unnecessary allocations when crossing WASM FFI boundary paritytech/substrate#10191
• Reduce overhead of generating network event metrics paritytech/substrate#10839
• Reduce CPU overhead of gossip paritytech/substrate#10859
• Further reduce the CPU overhead of networking metrics paritytech/substrate#10875
• Prevent double allocation of the payload when calling sp_io::storage::get paritytech/substrate#11523

A lot of PRs related to the WASM executor:

• Refactor WASM module instantiation paritytech/substrate#10480
• Add a new host function for reporting fatal errors; make WASM backtraces readable when printing out errors paritytech/substrate#10741
• Add extra WASM heap pages when precompiling the runtime blob paritytech/substrate#11107
• Catch panics on the FFI boundary between the runtime and the host for wasmtime paritytech/substrate#11189
• Improve wasmtime error reporting paritytech/substrate#11856
• Refactor sp-sandbox; make sure both sandbox executors are always tested paritytech/substrate#10173
• Clear WASM linear memory on other OSes besides Linux too when using fast instance reuse paritytech/substrate#10291
• Fix WASM executor without instance reuse; cleanups and refactoring paritytech/substrate#10313
• Statically register host WASM functions paritytech/substrate#10394

This includes a new instance reuse mechanism in wasmtime: bytecodealliance/wasmtime#3691; while they ultimately did not use my code directly what ended up in wasmtime was based on my work, where in this PR we switch to that new instance reuse mechanism in Substrate.

Extra software and hardware metrics:

• Add new hardware and software metrics paritytech/substrate#11062
• Add per-chain aggregate software/hardware telemetry paritytech/substrate-telemetry#464

Investigative work

Leaking memory? Something's crashing randomly? Are you wasting so much CPU time that you don't need any extra heating in winter? Who ya gonna call? Ghostbusters! Probably me!

Besides direct contributions to Substrate and Polkadot I also do a lot of investigative work actually finding the issues, mostly related to memory, performance and soundness. Some examples of that:

• For a memory leak crashing parachain nodes: my analysis and my PR with a fix
• For memory-related nondeterminism triggering a panic within the runtime: my analysis and my PR with a fix
• Fix for a memory leak crashing RPC nodes
• Found a dangerously long stack trace which was then fixed by Basti
• The contracts' benchmarks being up to 50% slower for no apparent reason; my analysis (part 1), my analysis (part 2)

I'm also the maintainer of Bytehound, a memory profiler for Linux, which I regularily use on Substrate/Polkadot when investigating issues.

Rank

Based on strict reading of the requirements as well as Gav's clarifications I apply for a rank of 1.

Updated to rank 2. (:

[bkchr]

Thinking about this again, I would probably approve II Dan for you.

Let's see what the others are thinking about it.

CC @jacogr @andresilva

[andresilva]

I would support rank 2 as well. I think @koute has proven deep knowledge on our WASM execution infrastructure and has done major contributions to improve it, also outside substrate (e.g. in wasmtime).

[rrtti]

@andresilva @bkchr do we want to merge as it is or ask @koute to change the rank from I to II?

[jacogr]

Based on the WASM execution infrastructure changes alone, I'm also in favor of a II Dan here, so can take my approval as-is if others concur. (My initial rating was already for somebody that should climb up one exceptionally quickly based on the work performed in this specific environment).

I also think the existing work can be turned into the basis of a strong "one published long-form semi-technical article concerning Polkadot" which should be attacked reasonably soon to conform to "This leniancy, such as it is, should exist only for the initial bootstrapping phase. By the first continuation grading, the full expectations of the member's rank should be met for approval by voters."

[koute]

Thank you all for the vote of confidence. (: I've updated the rank to 2.

[rrtti]

Hello @koute! Please let us know your Element handle to add you to the Fellowship channel. Thank you!

[koute]

Hello @koute! Please let us know your Element handle to add you to the Fellowship channel. Thank you!

Hey; it's @jan:matrix.parity.io