● Bash(cd ~/git/navicore/patch-seq && fj issue comment 475 -m "Scope narrowing — \`net.tcp.connect\` is done and #21

Merged

navicore merged 1 commit from new-tcp into main

2026-05-13 19:03:49 +00:00

navicore commented

2026-05-13 19:02:29 +00:00

Owner

  shipped (thank you!). What remains in this issue is th…)

⎿ Waiting…

──────────────────────────────────────────────────────────────────────────────────────────────────────────────────
Bash command

seq:allow(unchecked-tcp-write) on the connector/echo-server words. Acknowledges that the test is not the

place to teach write-failure handling, while keeping the lint useful elsewhere. Matches how the stdlib uses the
annotation for unchecked-modulo, deep-nesting, etc.
2. Ex 04 became peer-close detection, not "failure-path handling." After experimenting I confirmed net.tcp.read
on a peer-closed socket returns ("" true) — empty data with success — so the canonical TCP "they hung up"
detection is success && length == 0, not the Bool alone. That's a meatier and more honest lesson than a
contrived hard failure.
3. Fixed ports, not OS-assigned. Until issue #475's second half (returning the bound port from listen 0) lands,
we have to pick numbers. Used 18261-18265.

shipped (thank you!). What remains in this issue is th…) ⎿ Waiting… ────────────────────────────────────────────────────────────────────────────────────────────────────────────────── Bash command 1. # seq:allow(unchecked-tcp-write) on the connector/echo-server words. Acknowledges that the test is not the place to teach write-failure handling, while keeping the lint useful elsewhere. Matches how the stdlib uses the annotation for unchecked-modulo, deep-nesting, etc. 2. Ex 04 became peer-close detection, not "failure-path handling." After experimenting I confirmed net.tcp.read on a peer-closed socket returns ("" true) — empty data with success — so the canonical TCP "they hung up" detection is success && length == 0, not the Bool alone. That's a meatier and more honest lesson than a contrived hard failure. 3. Fixed ports, not OS-assigned. Until issue #475's second half (returning the bound port from listen 0) lands, we have to pick numbers. Used 18261-18265.

navicore added 1 commit

2026-05-13 19:02:29 +00:00

● Bash(cd ~/git/navicore/patch-seq && fj issue comment 475 -m "Scope narrowing — \net.tcp.connect\ is done and

CI - Linux / CI - Linux x86_64 (pull_request) Successful in 48s

Details

534134b790

shipped (thank you!). What remains in this issue is th…)
  ⎿  Waiting…

──────────────────────────────────────────────────────────────────────────────────────────────────────────────────
 Bash command
  1. # seq:allow(unchecked-tcp-write) on the connector/echo-server words. Acknowledges that the test is not the
  place to teach write-failure handling, while keeping the lint useful elsewhere. Matches how the stdlib uses the
  annotation for unchecked-modulo, deep-nesting, etc.
  2. Ex 04 became peer-close detection, not "failure-path handling." After experimenting I confirmed net.tcp.read
  on a peer-closed socket returns ("" true) — empty data with success — so the canonical TCP "they hung up"
  detection is success && length == 0, not the Bool alone. That's a meatier and more honest lesson than a
  contrived hard failure.
  3. Fixed ports, not OS-assigned. Until issue #475's second half (returning the bound port from listen 0) lands,
  we have to pick numbers. Used 18261-18265.

navicore merged commit 80500dd664 into main

2026-05-13 19:03:49 +00:00

navicore deleted branch new-tcp

2026-05-13 19:03:49 +00:00

navicore referenced this pull request from a commit

2026-05-13 19:03:50 +00:00

Merge pull request '● Bash(cd ~/git/navicore/patch-seq && fj issue comment 475 -m "Scope narrowing — \`net.tcp.connect\` is done and' (#21) from new-tcp into main