harden cli tests

[tao-stones]

Problem

TestValidator can be created with custom specified target_lamportds_per_signature

    pub fn with_custom_fees(
        mint_address: Pubkey,
        target_lamports_per_signature: u64,
        faucet_addr: Option<SocketAddr>,
        socket_addr_space: SocketAddrSpace,
    ) -> Self 

Cli tests uses with_custom_fees() but with target_lamports_per_signature == 1. This works accidentally because calculate_fee() currently uses fee_structure's default lamports_per_signature so long fee_rate_governor.lamport_per_signature <> 0.

Calculate_fee() is going to change (eg. be corrected). But in any case, cli tests should use correct value in setting up TestValidator.

Summary of Changes

• setting up TestValidator with correct target_lamports_per_signature

Fixes #

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Comparison is base (8f9d915) 81.8% compared to head (2dc88e2) 81.7%.
Report is 1 commits behind head on master.

Additional details and impacted files

@@            Coverage Diff            @@
##           master   #34844     +/-   ##
=========================================
- Coverage    81.8%    81.7%   -0.1%     
=========================================
  Files         825      825             
  Lines      223269   223269             
=========================================
- Hits       182635   182627      -8     
- Misses      40634    40642      +8     

[behzadnouri]

is this correct? it says 2 * fee_two_sig which would be 4?!

[tao-stones]

it's original test logic :) I only rename the fee to fee_two_sig for consistency. It checks 2 * fee_two_sig at ln 368 below, feels intended.

[tao-stones]

read the test, it sign a two-signature transaction twice that transfers 42.0 sol. offline_fee_payer_signer is one of the signer (eg payer) that is initiated with 1 + 2 * fee_two_sig. After submitted transaction, it's balance should be original_balance - fee_for_two_sig_transaction = (1 + 2 * fee_two_sig) - fee_two_sig = 1 + fee_two_sig. Which it asserts at Ln#470:

    check_balance!(
        sol_to_lamports(1.0) + fee_two_sig,
        &rpc_client,
        &offline_fee_payer_signer.pubkey(),
    );

So you are right that offline_fee_payer_signer does not need to be initialized with 2 *. But the test is still valid, even a bit of confusing. Happy to straight it out separately.