You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
#2881 reduces seq overhead in EVM - that said, @jangko the design where newComputation sometimes does database lookups feels pretty hairy - can't that be moved somewhere else? it feels like unnecessary maintenance overhead to do the same thing in two places
[20.23]
basically, for precompiles in particular, we don't execution evm opcodes so correspondingly we don't need a stack/memory - the natural way to solve that would be that the code is always resolved "outside" of newComputation and if code is nil, we skip stack/mem allocation instead of that ugly flag..
this also would help with the logic of reducing database lookups for precompiles - the database check is unnecessary (and expensive) - the PR introduces an ugly hack to fix this but ...
The text was updated successfully, but these errors were encountered:
From Discord chat:
#2881 reduces seq overhead in EVM - that said, @jangko the design where newComputation sometimes does database lookups feels pretty hairy - can't that be moved somewhere else? it feels like unnecessary maintenance overhead to do the same thing in two places
[20.23]
basically, for precompiles in particular, we don't execution evm opcodes so correspondingly we don't need a stack/memory - the natural way to solve that would be that the code is always resolved "outside" of newComputation and if code is nil, we skip stack/mem allocation instead of that ugly flag..
this also would help with the logic of reducing database lookups for precompiles - the database check is unnecessary (and expensive) - the PR introduces an ugly hack to fix this but ...
The text was updated successfully, but these errors were encountered: