Currently our benchmark script only updates the weights for production runtimes. For consistency, and to detect bugs caused by particular weight values, it should also update test runtime weights.
Steps to fix
- Add the test weights to the benchmark script
- Check in the script that all weights were regenerated, by searching for
DATE: and checking they're all the same
- Open a PR with the regenerated weights
Weights to fix
Special Cases
Needs to be manually checked first time, because it has no DATE:
Inconsistent module name:
The module name for weights_from_domains should also be renamed to match production:
|
weights::pallet_messenger_between_domains_extension::WeightInfo<Runtime>, |
Further work
Ideally, this test code would only be compiled when we're running tests, via a test-only feature.
We should also be only implementing these test weights for test runtimes, and not for all runtimes, so we can't accidentally use them incorrectly.
Currently our benchmark script only updates the weights for production runtimes. For consistency, and to detect bugs caused by particular weight values, it should also update test runtime weights.
Steps to fix
DATE:and checking they're all the sameWeights to fix
Special Cases
Needs to be manually checked first time, because it has no
DATE:Inconsistent module name:
The module name for
weights_from_domainsshould also be renamed to match production:subspace/domains/runtime/evm/src/lib.rs
Line 632 in 93b12c5
Further work
Ideally, this test code would only be compiled when we're running tests, via a test-only feature.
We should also be only implementing these test weights for test runtimes, and not for all runtimes, so we can't accidentally use them incorrectly.