The bytes crate is extremely popular for handling I/O and parsing buffers in the crates ecosystem, especially in respect to crates that build on Tokio.
It ends up being rather annoying/inefficient to use a bytes::Bytes instance with der types, as it either involves creating a wrapper struct, e.g.:
pub struct BytesAsOctetString(pub Bytes);
impl der::FixedTag for BytesAsOctetString {
const TAG: der::Tag = der::Tag::OctetString;
}
impl der::EncodeValue for BytesAsOctetString {
fn value_len(&self) -> der::Result<Length> {
OctetStringRef::new(&self.0)?.value_len()
}
fn encode_value(&self, encoder: &mut impl Writer) -> der::Result<()> {
OctetStringRef::new(&self.0)?.encode_value(encoder)
}
}
impl<'a> der::DecodeValue<'a> for BytesAsOctetString {
fn decode_value<R: Reader<'a>>(reader: &mut R, header: Header) -> der::Result<Self> {
Ok(BytesAsOctetString(
OctetString::decode_value(reader, header)?
.into_bytes()
.into(),
))
}
}or copying to/from an OctetString. This also invites potential mistakes because the most natural type to convert to is Vec<u8>, which der supports but which actually represents a SEQUENCE OF INTEGER, not an OCTET STRING.
In contrast, Bytes is pretty much as unambiguous as OctetString itself, with the caveat that it doesn't uphold the max-length guarantee that OctetString does.
This is an integration I'd be happy to contribute, because it would let me delete a good handful of lines of code on my end.
The
bytescrate is extremely popular for handling I/O and parsing buffers in the crates ecosystem, especially in respect to crates that build on Tokio.It ends up being rather annoying/inefficient to use a
bytes::Bytesinstance withdertypes, as it either involves creating a wrapper struct, e.g.:or copying to/from an
OctetString. This also invites potential mistakes because the most natural type to convert to isVec<u8>, whichdersupports but which actually represents aSEQUENCE OF INTEGER, not anOCTET STRING.In contrast,
Bytesis pretty much as unambiguous asOctetStringitself, with the caveat that it doesn't uphold the max-length guarantee thatOctetStringdoes.This is an integration I'd be happy to contribute, because it would let me delete a good handful of lines of code on my end.