Crate kinavis_nmea0183 

NMEA 0183 sentences to and from the KINAVIS types.

Anti-corruption layer between receiver sentences ($GPRMC,...*hh) and the domain:

1. Framing: length, printable characters, start character and a mandatory checksum. Any failure is an NmeaError.
2. Decoding into typed records — Rmc, Gga, Gll, Vtg — whose fields are kernel types (Position, Speed, TrueCourse); out-of-domain values (latitude 95°) and implausible ones (see Plausibility bounds) are rejected here. AIS arrives as Vdm with the payload still armoured; decoding it is the AIS crate’s job.
3. Translation into a GnssFix via TryFrom<Rmc> or Gga::fix_on.

Every step works on the input in place: no copies, no allocation, no panics. The crate builds for bare-metal targets without an allocator; CI checks the strict-profile build for panic paths.

encode and Display write a record back with its checksum, for generating sentences and for round-trip tests.

use kinavis_nmea0183::{parse, Sentence};
use kinavis_kernel::GnssFix;

let line = b"$GPRMC,225444.00,A,4916.4500,N,12311.1200,W,3.0,272.5,110926,5.0,W,D*30\r\n";
let Sentence::Rmc(rmc) = parse(line)? else { panic!("not an RMC") };

let fix = GnssFix::try_from(rmc)?;
assert_eq!(format!("{}", fix.taken_at()), "2026-09-11T22:54:44.000 UTC");
assert_eq!(format!("{:.2}", fix.position()), "49°16.45'N 123°11.12'W");
assert_eq!(fix.speed_over_ground().map(|s| s.knots()), Some(3.0));
assert!(fix.quality().fix_type().is_position_fix());

// And back out again, checksum recomputed.
assert_eq!(format!("{rmc}"), "$GPRMC,225444.00,A,4916.4500,N,12311.1200,W,3.0,272.5,110926,5.0,W,D*30");

Plausibility bounds

A value outside these inclusive bounds is a corrupt field, not a measurement, and fails with NmeaError::Value:

Field	Bounds
Speed over ground	0 to 1000 kn; 0 to 1852 km/h
Altitude (GGA)	−10 000 m to 100 000 m
Geoid separation (GGA)	−1000 m to 1000 m
Dilution of precision	0 to 100; 0 reads as not available
Magnetic variation (RMC)	180° either way; 999 and above read as not available
Differential age (GGA)	0 s to 9999 s

parse accepts sentences up to MAX_ACCEPTED_BYTES, past the standard’s MAX_SENTENCE_BYTES: receivers write longer ones. encode refuses a sentence longer than MAX_SENTENCE_BYTES with NmeaError::TooLong. RMC, GLL and VTG always fit; a VDM fits with at most 62 payload characters; a GGA can exceed the limit only with several fields near their bounds at once.

Not supported

Other sentences (satellites in view, DOP breakdowns, proprietary) return Sentence::Unsupported with their address and a verified checksum, for counting or logging.

Feature flags

• std (default) — standard library maths in the kernel.
• libm — for no_std targets: --no-default-features --features libm.

Structs

Address

Sentence address field; at most 10 characters (proprietary maximum).

Date

Date as carried by a sentence.

Gga

Fix data: position, time, quality, satellites, HDOP, altitude.

Gll

Geographic position: latitude, longitude, time.

Payload

Armoured payload of one fragment.

Rmc

Recommended minimum: position, velocity, time and date.

Talker

Two-letter talker identifier: GP GPS, GN multi-constellation, GL GLONASS, etc.

TimeOfDay

UTC time of day as carried by a sentence, at the talker’s resolution.

Vdm

AIS message or one fragment of it.

Vtg

Course and speed over the ground.

Enums

Channel

AIS radio channel.

Mode

Mode indicator (NMEA 2.3+): how the position was obtained.

NmeaError

Why a sentence could not be read or written.

Sentence

Parsed sentence, supported or not.

Status

Data status of a position sentence: A or V.

TranslationError

Why a well-formed sentence could not become a GnssFix.

Constants

MAX_ACCEPTED_BYTES

Maximum sentence length parse accepts, $ and CR LF included.

MAX_FRAGMENTS

Maximum number of fragments per message.

MAX_PAYLOAD_CHARS

Maximum payload characters per fragment.

MAX_SENTENCE_BYTES

Maximum sentence length in the standard, $ and CR LF included. encode never writes more.

Functions

encode

Writes a sentence ($ to CR LF) into out and returns its length.

parse

Parses one sentence.