buffalojoec

Loader V4 Redux

After discussions on RFC-0423, it's clear that most contributors prefer to introduce the new BPF Loader account layout under a new program ID. Therefore, we plan to proceed with a BPF Loader V4.

I'd like to consider a new direction on the design for Loader V4. SIMD-0167 is well-specified and addresses several developer pain points. However, I believe it oversteps in some areas and falls short in others.

buffalojoec

use {
    bytemuck::{Pod, Zeroable},
    pinocchio::{
        account_info::AccountInfo, program_entrypoint,
        pubkey::Pubkey, ProgramResult, program_error::ProgramError,
    },
};

#[derive(Clone, Copy, Pod, Zeroable)]
#[repr(C)]

buffalojoec

use solana_program::{account_info::AccountInfo, entrypoint::ProgramResult, msg, pubkey::Pubkey};

solana_program::entrypoint!(process);

fn process(_program_id: &Pubkey, accounts: &[AccountInfo], _input: &[u8]) -> ProgramResult {
    let account = accounts.first().unwrap();
    msg!("KEY:          {}", account.key);
    msg!("IS_SIGNER:    {}", account.is_signer);
    msg!("IS_WRITABLE:  {}", account.is_writable);
    Ok(())

buffalojoec

//! Program cache index v2 implementation.
//!
//! Consider a fork graph.
//!
//! ```
//!
//!          0 <- root
//!        /   \
//!       10    5
//!       |     |

buffalojoec

runtime::bank::new_from_fields()
  |
  |-- runtime::serde_snapshot::reconstruct_bank_from_fields()
      |
      |-- runtime::serde_snapshot::bank_from_streams()
          |
          |-- runtime::snapshot_bank_utils::rebuild_bank_from_unarchived_snapshots()
          |   |
          |   |-- runtime::snapshot_bank_utils::bank_from_snapshot_archives()
          |       |

buffalojoec

/// Scenario #1:
mod not_fut{
    /// A lib function designed to take a closure that returns a `String`
    fn lib_function<F>(get_account_data_fn: F) -> String
    where
        F: Fn(u8) -> String,
    {
        let account_data = get_account_data_fn(42);
        println!("Account data: {}", account_data);
        account_data