Sixteen Means Unreachable

Core Question

Core Question

RIP needs a metric value that fits in the same integer field as every real distance but means "not reachable at all." Why is that value 16 — not 255, not a billion — and what does picking a small number actually buy?

Outcome

Outcome

By the end of this session, the learner should be able to:

• state INFINITY = 16 and explain why the value is deliberately small rather than deliberately large
• give the exact boundary condition is_unreachable() uses, and evaluate it at 15 and at 16
• explain what clamp_metric() does to a sum that overshoots 16, with a concrete example
• state what poison() changes about a route and what it leaves untouched

Read Order

Read Order

1. Read the module docstring
2. Read INFINITY
3. Read is_unreachable()
4. Read clamp_metric()
5. Read poison()
6. Run examples/rip/session_03_walkthrough.py

Read It Like Code

Read It Like Code

is_unreachable(metric)   # metric >= INFINITY
clamp_metric(metric)     # min(metric, INFINITY)
poison(route)            # replace(route, metric=INFINITY)

Fields That Matter

Fields That Matter

There is no dataclass introduced in this module — infinity.py is three small functions and one constant, operating on plain integers and on the RipRoute defined in session 01.

Decision Flow

Decision Flow

is_unreachable(metric):
  metric >= INFINITY   -> True
  otherwise             -> False

clamp_metric(metric):
  metric > INFINITY    -> INFINITY
  otherwise             -> metric   (min() collapses both cases into one expression)

poison(route):
  -> RipRoute(route.prefix, INFINITY, route.next_hop, route.learned_from)
     (every field but metric is copied unchanged)

Reading Lens

Reading Lens

This module is pure and tiny in exactly the way tcp/seqnum.py is pure and tiny: no dataclasses to inspect, no branching state machine, just arithmetic that has to be exactly right at one boundary. The important move is to stop treating "16" as an arbitrary magic number and start asking:

• what specific comparison operator does is_unreachable() use at the boundary — is 16 itself unreachable, or is 17 the first unreachable value?
• when clamp_metric() fires, what real-world event does an overshoot like 15 + 1 = 16 or advertised 100 represent?
• poison() builds a new RipRoute with one field changed. Which fields does it have to carry forward unchanged for the result to still mean "this is a route to the same destination, just now marked dead"?

Toy Model Boundary

Toy Model Boundary

The module docstring itself sets up the next session: a small infinity bounds how many rounds a routing loop can circulate before the bad rumor tops out, at the cost of also bounding how large a real network RIP can cover (15 hops maximum, since 16 means unreachable). This module only implements the boundary check and the poisoning operation — it does not implement the loop, the timers, or the periodic-update behavior that actually drives a metric upward round by round. That is count_to_infinity.py's job, in the session this one is deliberately positioned before.

This module's specific numbers — INFINITY = 16, and the discussion of hop-count semantics — are RIPv2's. RIPng (RFC 2080) reuses the same metric field and the same value of 16 for IPv6, but carries it differently on the wire; this module does not model that difference.

Code Landmarks

Code Landmarks

Module docstring

Frames 16 not as a limit on network size so much as a deliberately short fuse on how long a bad rumor can circulate — a design tradeoff, not an accident. Read this before the functions below; it is the argument the Failure Questions test.

is_unreachable()

One comparison, metric >= INFINITY. Note the >=, not > — a metric of exactly 16 already counts as unreachable, it does not have to exceed 16 first.

clamp_metric()

min(metric, INFINITY) — a single expression that leaves any metric of 16 or below untouched and caps anything above it at exactly 16. This is the function session 02's process_advertisement() calls on every candidate before it ever compares metrics, so no route in the table can ever silently carry a metric above 16.

poison()

dataclasses.replace(route, metric=INFINITY) — the only field touched is metric; prefix, next_hop, and learned_from all pass through unchanged. Poisoning a route does not erase who it came from, it just declares the distance to it infinite.

Failure Questions

Failure Questions

Use the source file to answer these:

1. is_unreachable(16) and is_unreachable(15) — what does each return, and which operator (>= versus >) makes 16 itself count as unreachable rather than the last reachable hop count?
2. clamp_metric(15 + 1) and clamp_metric(100) both return the same value. What does that tell you about how much information clamp_metric() preserves about *how far* a metric overshot infinity?
3. poison() uses dataclasses.replace() rather than constructing a new RipRoute field by field. What has to be true about RipRoute (see session 01) for replace() to be usable at all here?
4. After poison(route), is poisoned.learned_from == route.learned_from? Read the replace() call to justify your answer — does poisoning a route change who it was learned from?
5. The module docstring says a small infinity is "a deliberately short fuse," not a network-size limit as an end in itself. If INFINITY were 256 instead of 16, what specifically about is_unreachable() and clamp_metric() would still work correctly, and what would only change in degree (how long a bad rumor could circulate) rather than in kind?

Walkthrough

Walkthrough

Run this:

PYTHONPATH=src python3 examples/rip/session_03_walkthrough.py

The walkthrough clamps 15 + 1 and a wildly overshot 100 to confirm both land on exactly 16, checks the is_unreachable() boundary at 16 and at 15, and poisons a route to confirm the metric becomes 16 while prefix, next_hop, and learned_from all survive unchanged.

Done When

Done When

The learner can say all of the following without looking at notes:

• "16 is INFINITY — small on purpose, so a routing loop's metric tops out in a handful of rounds instead of climbing forever."
• "is_unreachable() uses >=, so a metric of exactly 16 is already unreachable, not the last valid hop count."
• "poison() only ever changes metric; every other field of the route — including who it was learned from — is carried forward untouched."

References

References

• RFC 2453 Section 3.8 (Distance Metric: RIP's use of hop count and the choice of 16 as infinity)
• RFC 2453 Section 3.10 (Timers, referenced here only to note what this toy model deliberately omits — the periodic and triggered-update timing that actually drives count-to-infinity)

Continue

Keep reading RIP