[LST-343] StartRebalance and EndRebalance instructions #89
Conversation
billythedummy
left a comment
billythedummy
left a comment
There was a problem hiding this comment.
i think theres also some confusion between Pubkey and [u8; 32]
Pubkeyis just a simple newtype of[u8; 32]and you can always convert from one to the other easily- we're only using
Pubkeybecause the solana dev deps (mollusk-svm,solana-instructionetc) use it. Otherwise, just stick with[u8; 32]
| // Filter out executable accounts (programs and sysvars) which are not | ||
| // relevant for rent-exemption checks - only user accounts matter | ||
| let mut result_for_check = result.clone(); | ||
| result_for_check | ||
| .resulting_accounts | ||
| .retain(|(_, acc)| !acc.executable); |
There was a problem hiding this comment.
which program account is not rent-exempt tho?
There was a problem hiding this comment.
Without the filter i would get this #89 (comment)
There was a problem hiding this comment.
#89 (comment) seems like it affects the balanced check, not the rent exemption check?
There was a problem hiding this comment.
Token Program starts with 934M lamports from fixtures in accs_bef, but after Mollusk executes the tx it appears in resulting_accounts with 0 lamports (token program account is reset in tests here? is that expected?). so we need to filter it out here before we do rent-exempt check
There was a problem hiding this comment.
no this isnt rly expected lol wtf, guess we can check some other simpler tests that involve the token program to see whats going on
There was a problem hiding this comment.
this doesnt happen in withdraw_protocol_fees_test_correct_basic. Im guessing theres somewhere where you've mutated bef or aft and youre not checking the actual input/output to svm.process_instruction_(chain). Or maybe mollusk is weird and only outputs the accounts that the last instruction involves in InstructionResult
There was a problem hiding this comment.
resulting_accounts: The resulting accounts after the last instruction.
https://docs.rs/mollusk-svm/latest/mollusk_svm/struct.Mollusk.html#method.process_instruction_chain
There was a problem hiding this comment.
yeah i thought so too but if you look at the impl for process_instruction (_chain just calls this in a loop): the resulting accounts is always whatever was input to the function, not just the ones that are involved in the instruction (theres no filter in sight): https://github.com/anza-xyz/mollusk/blob/8029da5304e78fcf1d4396515f1e43422ad550a9/harness/src/lib.rs#L776-L806
There was a problem hiding this comment.
seems like this is happening on the second token transfer instruction
billythedummy
left a comment
billythedummy
left a comment
There was a problem hiding this comment.
lets just do the <acc>_checked functions and then do other cleanups in subsequent PRs
| // Filter out executable accounts (programs and sysvars) which are not | ||
| // relevant for rent-exemption checks - only user accounts matter | ||
| let mut result_for_check = result.clone(); | ||
| result_for_check | ||
| .resulting_accounts | ||
| .retain(|(_, acc)| !acc.executable); |
There was a problem hiding this comment.
no this isnt rly expected lol wtf, guess we can check some other simpler tests that involve the token program to see whats going on
|
|
||
| const _ACC_DATA_ALIGN: usize = 8; | ||
|
|
||
| const _POOL_STATE_ALIGN_CHECK: () = assert!(core::mem::align_of::<PoolState>() <= _ACC_DATA_ALIGN); |
There was a problem hiding this comment.
whyd you delete this?
There was a problem hiding this comment.
happy little accident
Johnnycus
deleted the
lst-343-inf1-jiminy-reimpl-startrebalance-and-endrebalance
branch
This PR contains the
StartRebalanceandEndRebalanceinstructions reimplementation with jiminy.