Closed crystalin closed 2 years ago
@crystalin if that is failing reproducible. You will need to write some script that goes back on your master until the tool works again. If you have his this point, you know which commit to look at. I would not assume that something broke there. In Substrate & Polkadot we have CI jobs running that ensure that two build runs produce the same wasm binaries. Before we had achieved this, we also needed to fix some stuff.
@bkchr Thanks, I'm going to track the regression and ping you if I can't identify the reason
@bkchr I confirm it is when upgrading from v0.9.12 to v0.9.13 I'm looking at the PRs, but if you have any idea where that could be
You can just write a simple script to use with git bisect. Then the computer can do this work for you 😬
This was caused by using the crate gensym
to generate some module names in our codebase, fixed here:
https://github.com/PureStake/moonbeam/pull/1328
I think this can be closed.
Nice! Ty for reporting back. If you want to prevent this, you could copy the reproducible wasm build CI check from Substrate :)
Something must have changed in substrate recently I suppose that is breaking the deterministic part of srtool. Ex:
Will generate