Interesting !
Got some concerns though

1. This feature makes the code more complex, can you quantify how much gas this can save during pre-liquidation deployment ? Want to make sure that this is worth it
2. Implementing transient storage requires either to update the solidity version to 0.8.28 to enable the use of the transient keyword either to use assembly to use TLOAD and TSTORE, potentially making the code harder to read
3. If we were to implement this, would we need to re-audit the whole project ? From a security standpoint it seems like a minor update but it still means using new opcodes so it would be best to run another round of audit

This feature makes the code more complex, can you quantify how much gas this can safe during pre-liquidation deployment ? Want to make sure that this is worth it

Not sure but this should be less than 20k gas, the cost of a very cold SSTORE. This may not be the most important benefit of this PR though: making it possible to have deterministic addresses onchain, and being able to have a nice error message should bring the most benefit

Agree with point 2, until we get proper support in solidity.

Also agree with point 3, this is probably too late. But this was more meant as a demonstration of what we could do in the future than an actual PR we should merge right now

• do you have a gas cost comparison?
• note that the deployment of a preliquidation is approximately 1.4M gas anyway
• yes

do you have a gas cost comparison?

No, because the PR is unusable as of now (see point 2 of the todo list). I will close it, as it was more something to have in mind for the future than anything else