Frame

Lesson 202 established that storage density directly determines range — how far back the ring can see. The ring is bounded. Multiple modalities compete for the same space. Every byte wasted on format overhead is a byte that could hold another imprint. Every lost imprint shortens the range. Less range, less knowledge. The frame must be the densest possible container for the imprint. This lesson derives what that container looks like.

Density

This is not a preference. This is a constraint. The frame — the container that holds the imprint — must be the densest possible representation of what the sensor captured. Not approximately dense. Not "good enough." The densest achievable. Because in a bounded ring, density is range, and range determines how much the corpus knows.

The question is: what does a container look like when maximum density is a hard requirement?

No names on the wire

A text format carries the name of every field in every frame. "temperature": 72.4. "timestamp": 1713225600. The word "temperature" is eleven bytes. The value is four. The name costs nearly three times what the data costs. Multiply that across thousands of imprints on the ring and the overhead consumes more space than the witnesses.

The first principle: no names on the wire. The writer and the reader both know the schema in advance — what fields exist, in what order, at what size. The schema was declared before the first imprint was ever captured. If both sides know the shape, the frame does not need to describe itself field by field. It carries the data and nothing else.

The type code — established in Lesson 202 — identifies the modality. The modality identifies the schema. The schema tells you what each byte means. No field names. No delimiters. No overhead. The frame is pure substance.

Presence, not position

Not every imprint of the same modality carries the same fields. A temperature reading might include humidity this time and not the next. A system observation might include memory pressure on one tick and omit it on another. The sensor reports what it sensed. If a field was not sensed, it should not occupy space.

A rigid layout — every field at a fixed byte offset, always present, always the same size — wastes space on empty fields. That violates the density requirement. But a fully flexible format — fields in any order, self-describing, variable — wastes space on description. That also violates the density requirement.

The answer is in between. A fixed header carries a flags word. Each bit in the flags word says whether a particular slot is present or absent. The slots that are present pack contiguously after the header. No gaps. No empty fields. No wasted bytes. The flags word is the guide — the reader checks the flags to know what is there and where each slot begins.

This gives density and variation simultaneously. The frame holds only what was actually sensed. Nothing is padded. Nothing is wasted. And the reader does not parse — it reads the flags, computes the offset, and goes directly to the data. Access is still a pointer operation, not a traversal.

Fixed byte order

The frame may be written on one machine and read on another. It may be written today and read years from now, on hardware that does not exist yet. The bytes must mean the same thing everywhere, always, regardless of who reads them or when.

This requires a canonical byte order. One endianness. Declared once, respected everywhere. The frame does not negotiate byte order at read time. The frame does not carry a byte-order marker. Every multi-byte field is stored in the declared order. Period.

Canonical byte order is preservation made physical. Lesson 104 established four requirements for preservation: integrity, reconstructability, governance, durability. A frame with canonical byte order satisfies all four. The imprint is unaltered (integrity). Any reader can reconstruct the observation accurately (reconstructability). The byte order is explicitly defined (governance). The frame does not depend on a particular runtime or architecture (durability).

Variations

One corpus holds many modalities. Temperature, acceleration, GPS, camera, heartbeat. Each modality has a different schema — different fields, different sizes, different byte counts. The frames on the ring are not all the same size. They are not all the same shape.

The frame must handle this. Each frame carries its type code and its length. The type code says what modality. The length says how many bytes. Together they let any reader walk the ring — skip what it does not need, stop at what it does. Different modalities coexist on the same ring without interfering with each other.

A small frame — a temperature reading, a few bytes — sits next to a large frame — a camera snapshot, hundreds of bytes. Both are valid. Both are first-class. The ring does not care about the size. The ring cares about the space. Every frame earns its space by carrying an imprint that deserves preservation.

The canonical ring

Assemble the pieces.

The ring from Lesson 105 — the circular buffer derived from first principles — holds frames. Each frame captures a moment of reality. Each frame is packed to maximum density — no names, presence-flagged slots, canonical byte order. Each frame identifies its modality and its length. The ring holds them all, in the order they arrived, on a shared timeline.

The ring is the canonical source.

What is on the ring is truth. The frames are packed canonically — same layout, same byte order, same schema, always. Any reader, on any machine, at any time, can pick up a frame and read it without negotiation. The ring does not adapt to the reader. The reader adapts to the ring. Because the ring holds the canonical form — the one true representation of the imprint as it was captured.

A canonical ring holding frames of time with their respective imprints. That is the physical form of the corpus at the storage layer. The principles from the 100 level — circular structure, temporal depth, preservation — made concrete in packed, canonical, typed frames on a bounded ring.