Struct SubgroupHeader

Source

pub struct SubgroupHeader {
    pub header_type: u8,
    pub track_alias: VarInt,
    pub group_id: VarInt,
    pub subgroup_id: VarInt,
    pub publisher_priority: Option<u8>,
}

Fields§

§header_type: u8§track_alias: VarInt§group_id: VarInt§subgroup_id: VarInt§publisher_priority: Option<u8>

Implementations§

Source §

impl SubgroupHeader

Source

pub fn has_extensions(&self) -> bool

Source

pub fn subgroup_id_from_first_object(&self) -> bool

Whether SUBGROUP_ID_MODE is 1: the Subgroup ID is the first object’s Object ID and is not transmitted on the wire.

Reads the two bits as the one field Section 10.4.2 defines, rather than as the single bit 0x02. The single-bit reading answered true for a Type naming the reserved mode 3, where 0x02 and 0x04 are both set — so this predicate and Self::has_explicit_subgroup_id answered true together, for a state one two-bit field cannot be in.

Source

pub fn has_explicit_subgroup_id(&self) -> bool

Whether SUBGROUP_ID_MODE is 2: an explicit Subgroup ID field follows the Group ID.

Self::decode refuses the reserved mode before reading any field, so on a decoded header this agrees with the single-bit 0x04 reading it replaces. The two differ only on a header built by hand, which is the only way to hold a reserved-mode Type at all.

Source

pub fn has_end_of_group(&self) -> bool

Source

pub fn has_priority(&self) -> bool

Source

pub fn encode(&self, buf: &mut impl BufMut)

Serialize the header exactly as its Type byte describes it.

Infallible, and so willing to write a Type draft-16 Section 10.4.2 tells an endpoint to reject — including a reserved-mode Type this module’s own Self::decode refuses to read back. Prefer Self::encode_checked, which refuses those Types instead.

A reserved-mode Type gets no Subgroup ID field, because mode 3 defines none: the draft says what the other three modes put on the wire and nothing about this one. Writing the field there was an artifact of reading 0x04 on its own, and it disagreed with every neighbouring draft — 15, 17, 18 and 19 all leave it off.

Every field is driven by the Type byte, because that is the only thing Self::decode and the peer have to go on — the Priority byte included. Driving that one off publisher_priority being Some instead desyncs the stream: a Type with DEFAULT_PRIORITY (0x20) set beside a Some writes a stray byte the reader takes for the first Object’s Object ID Delta, and a Type with the bit clear beside a None leaves the reader taking that Delta for the priority. Either way every later field shifts — which is precisely the disagreement Self::encode_checked’s doc comment says this pair must not have.

A None under a Type whose bit is clear therefore writes 128 rather than dropping the byte. Draft-16 Section 11.1.1.1: “A subscription has Publisher Priorty 128 if this extension is omitted”, so 128 is this draft’s own name for an unstated priority and not an invented filler. Drafts 17-21 write the same value in the same place. Use Self::encode_checked to be told about the disagreement rather than having it resolved silently.

Source

pub fn encode_checked(&self, buf: &mut impl BufMut) -> Result<(), CodecError>

Serialize the header, refusing a Type value draft-16 forbids.

Errors with CodecError::InvalidField for exactly the Types Section 10.4.2 lists as invalid — the same set Self::decode refuses — before any byte is written, so a refused header leaves buf untouched.

The check belongs on this side as well because the two halves would otherwise disagree about which streams exist: a reserved-mode header written by Self::encode cannot be read back by Self::decode, and a codec used to rewrite captured traffic would emit a stream it could not then parse.

That argument covers the Priority byte too. publisher_priority disagreeing with the Type byte’s DEFAULT_PRIORITY bit (0x20) is the same failure one field along: a header that does not mean what it says, whose bytes the peer reads under framing its writer never asked for. Self::encode does not desync the stream over it — the byte follows the Type byte — but resolving the disagreement is not the same as being told about it, and the caller who set the wrong half is the only one who can fix it. Same check, same reason, as draft-15 Section 10.4.2’s encoder.

Source

pub fn decode(buf: &mut impl Buf) -> Result<Self, CodecError>

Decode a subgroup header, Type field included.

A Type spelled in more than one byte is refused before it is narrowed, by wide_type_refusal. Every Type draft-16 assigns fits a single byte, so a wider spelling is either a Type this draft does not have or a non-minimal spelling of one it does, and both have to be refused rather than truncated into a Type that happens to be valid.