Warnings

RUSTSEC-2023-0033

Parsing borsh messages with ZST which are not-copy/clone is unsound

Affected versions of borsh cause undefined behavior when zero-sized-types (ZST)
are parsed and the Copy/Clone traits are not implemented/derived.
For instance if 1000 instances of a ZST are deserialized, and the ZST is not copy
(this can be achieved through a singleton), then accessing/writing to deserialized
data will cause a segmentation fault.

There is currently no way for borsh to read data without also providing a Rust type.
Therefore, if not ZST are used for serialization, then you are not affected by this issue.

RUSTSEC-2023-0055

Multiple soundness issues

lexical contains multiple soundness issues:

1. Bytes::read() allows creating instances of types with invalid bit patterns
2. BytesIter::read() advances iterators out of bounds
3. The BytesIter trait has safety invariants but is public and not marked unsafe
4. write_float() calls MaybeUninit::assume_init() on uninitialized data, which is is not allowed by the Rust abstract machine

The crate also has some correctness issues and appears to be unmaintained.

Alternatives

For quickly parsing floating-point numbers third-party crates are no longer needed. A fast float parsing algorith by the author of lexical has been merged into libcore.

For quickly parsing integers, consider atoi and btoi crates (100% safe code). atoi_radix10 provides even faster parsing, but only with -C target-cpu=native, and at the cost of some unsafe.

For formatting integers in a #[no_std] context consider the numtoa crate.

For working with big numbers consider num-bigint and num-traits.