I am opening this technical thread before any revised on-chain submission, following feedback on proposal #12238.
The original proposal asks to add my wallet to code_upload_access. I understand the concern: the SHA-256 written in a proposal description is not enforced by that allowlist message, and the permission would remain usable until governance changes it. This thread is not a request to change votes on proposal #12238.
Public contract source and build recipe:
The public repository contains only the Avatar Club 100 CosmWasm contract package, tests, and build workflow; the RD Station dApp source remains private. The Wasm file I intend to submit has SHA-256:
6e2075f95c52b69b116e681d2a8b19778081f2b9de67819bb4063816ce120a84
The proposed safer path is a governance-authorized MsgStoreCode that stores this Wasm, so no individual wallet receives ongoing upload permission. Instantiation would be a separate later transaction. I am asking core developers and validators to review the source, reproducible build, exact message format, and any chain-specific requirements before I prepare a resubmission.
The repository states the expected hash and CI is currently verifying the build. I will add the successful run and artifact details once that check completes. No independent security audit has been performed, and I am not claiming that it has.
Please point out any technical issue or missing evidence needed for a governance MsgStoreCode proposal.
CI update: the public workflow completed successfully. It ran the contract formatting check and tests, built the Wasm, confirmed SHA-256 6e2075f95c52b69b116e681d2a8b19778081f2b9de67819bb4063816ce120a84, ran cosmwasm-check, rebuilt independently, and confirmed the two Wasm files are byte-for-byte identical. The workflow also uploaded the Wasm and checksum artifact.
The result is available in the public repository under Actions: successful run #2, “Pin Rust toolchain for reproducible build” (commit fa51539).
This verifies reproducibility for this public source snapshot and build environment; it is not an independent security audit.
I am able to run the build using the toolchain specified in the README.md and reproducible-build.yml and verify that the unit tests pass, but the built Wasm does not calculate to the hash you specified. The build I ran produced SHA-256 05813b28a6aed85b3e16d86dcd5b7461dd74ea4698617f67d2d27e24d60c388a
I found a concrete source of build-hash variation. Your script uses -Z build-std, which recompiles Rust’s standard library. The resulting Wasm embeds absolute paths to the installed Rust source files.
I tested this by rebuilding the unchanged contract with the same Rust 1.85.1 toolchain copied to a different absolute path. Moving the contract source directory alone produced identical bytes; moving the toolchain changed the hash from 05813b28a6aed85b3e16d86dcd5b7461dd74ea4698617f67d2d27e24d60c388a to 8e28b403ef40273105a2605fb256e7e38644d8180909a8a394d12f3823838134. I confirmed that the second Wasm embeds the new toolchain path.
This proves the current build is sensitive to the Rust toolchain installation path. It does not yet prove that this explains your published hash, 6e2075f95c52b69b116e681d2a8b19778081f2b9de67819bb4063816ce120a84.
Could you provide the Wasm file with that published hash and the build log, including the output of rustc +1.85.1 --print sysroot? That would let reviewers inspect the exact binary intended for the MsgStoreCode proposal. It may also be worth normalizing embedded source paths in the build process and publishing the resulting hash.
Thanks for publishing the source and opening this thread before resubmitting. That’s the right path.
Like Raider in post #5, we can’t treat the build as reproducible yet. The script compiles with -Z build-std, which writes the Rust toolchain’s install path into the Wasm, so independent builds give different hashes (Raider got 05813b…388a, not 6e2075…0a84). A CI run that passes on your own setup isn’t independent verification.
Before we could support a store-code proposal, we’d want to see:
A build anyone can reproduce to the same hash, ideally with the standard cosmwasm/optimizer image (or with paths normalized), and that hash published in the repo.
The exact Wasm and the full build log attached to a tagged release, so reviewers check the actual bytes that would be stored.
A clear explanation of the off-chain signed fee quotes: who holds the signing key, what happens if it leaks or is lost, and who can pause the contract or change its settings.
A new proposal that stores only that Wasm through governance (MsgStoreCode with the gov module as authority), with no wallet added to the upload whitelist.
Once those are in place we’re happy to review again. Since there’s no audit, a second independent reviewer would help too.