Crate rta_derive 

rta-derive

Procedural macro implementation for #[derive(RTA)]

#[derive(RTA)]

It generates an implementation of the rta::RTA trait for a struct T.

At compile time it computes:

• SIZE: core::mem::size_of::<T>()
• HASH: a deterministic u64 layout fingerprint

Hash ([T::HASH])

The [T::HASH] is computed from:

• the name of T
• the ordered list of field names
• the ordered list of field types

[T::HASH] is used to detcect type identity & schema changes for the type T

[!CAUTION] The [T::HASH] is not cryptographically secure, and is not intended to be ;). It is only used as a deterministic layout fingerprint.

Compile-time Checks

RTA structs are intended to be,

• read without serialization
• checksummed using raw byte access
• stored directly in memory-mapped files

These features requires RTA type to be,

• padding-free
• deterministic
• byte-stable across compilations

To achieve these, the #[derive(RTA)] macro performs following compile-time checks:

• T uses #[repr(C)]
• T is not zero-sized
• T contains no padding
• size_of::<T>() is multiple of 8 bytes

If any of these requirements are violated the compilation fails ;-)

[!NOTE] We require size_of::<T>() to be multiple of 8 bytes, this is done for better performance while calculating Cyclic Redundency Check (CRC)

Safety Requirements

Not all constraints can be enforced by compile time checks, these must be followed by users:

• T must not implement Drop
• fields should be plain data (POD-like)
• T must be Sized, and implement Default

Violating these rules may cause undefined behavior when the struct is stored directly in memory-mapped storage.

Derive Macros

RTA

Derives the rta::RTA trait for a struct T