Collaboration diagram for The flatlogs log protocol:

Modules
	Type definitions for flatlogs.

	The flatlogs binary header definition

	Standard text formatting of logs for biological purposes

	Flatlogs_files

Detailed Description

flatlogs consist of a simple header, followed by a flatbuffers encoded message payload.

The log entry is a binary buffer with the following format:

                     1 1 1 1 1 1 1 1 1 1 2 2 2 2 2 2 2 2 2 
 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8
|p|evt|time_s |time_ns|l| [message (0-254)]
|p|evt|time_s |time_ns|E|len| [message (255-65534)]
|p|evt|time_s |time_ns|F|len          | [message (>65535)]

where the header consists of everything up to the [message]. Table notations are:

p (uint8_t) denotes the log priority.
evt (uint16_t) denotes the event code.
time_s (uint32_t) is the time in seconds since the epoch
time_ns (uint32_t) denotes the nanoseconds past that second.
l/len is the message length field, which is variable in length depending on the size of the message:
- Message length 0-253: uint8_t in a single byte.
- Message length 255-65534: uint8_t = 0xFF-1 in the first byte, uint16_t in the next two bytes.
- Message length 65535 and up: uint8_t=0xFF in the first byte, uint64_t in the next 8 bytes.

Rationale for variable length: this keeps the space used for 0 length and short messages to a minimum, while allowing for progressively larger messages to be saved.
The savings will be large for small messages, while the the cost is proportionally small for larger messages. For larger messages, this requires a progressive read to sniff out the entire length. Likewise, the cost is small relative to the cost of reading the larger message.

If a non-zero message exists, it is usually a flatlogs serialized buffer.

Modules

Detailed Description