julian.kiedaischandClaude Opus 5
26ed1495fc
The dashboard's fifth guard learned to see a host name and found three in tools/live: a default host in lib.ts and two health checks written out. A measuring tool that aims at somebody else's server by default is a trap, so FILIUS_HOST has no fallback and the run says what to set; the two checks use BASE, which is derived from it. BASE had to be imported in m3.ts — without that this would have been a NameError instead of the intended refusal. README.md keeps the real names: the project is this school's and the prose describes its deployment; what has to stand on its own is the code. The allow file also excuses the simulator's own data. Filius simulates a network, so its scenarios, tests and documentation carry 65 693 addresses (measured) — all of invented machines in invented networks, which is the point of the program. Deliberately broad: a list of scenario files would be one nobody keeps correct. typecheck green. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
34 lines
1.7 KiB
Plaintext
34 lines
1.7 KiB
Plaintext
# Where an address or a host name of the school stands on purpose in this
|
|
# submodule. Applies **on top of** the dashboard's own list — the checker
|
|
# reads both (INTERNAL_ADDRESSES_ROOT points it here).
|
|
#
|
|
# <path>::<rule> # why
|
|
# host:<name>::<rule> # why, for a host that is excused everywhere
|
|
#
|
|
# Why each rule exists:
|
|
# INTERNAL_ADDRESSES_ROOT=$PWD python3 ../scripts/check_no_internal_addresses.py --rules
|
|
|
|
# ── Documentation stays specific to this school ─────────────────────
|
|
# Decided on 13.11.2026: the project *is* this school's, and the prose
|
|
# describes its deployment. What has to stand on its own in a dev container
|
|
# is the **code** — not the text about it. So the README keeps the real
|
|
# names while the configuration and the tools no longer carry any.
|
|
README.md::school-hostname
|
|
|
|
# ── This program's subject *is* networks ────────────────────────────
|
|
# Filius simulates a network, so every scenario, every test and every piece
|
|
# of documentation is full of addresses — 65 693 of them, measured on
|
|
# 13.11.2026. None belongs to a school: they are the addresses of invented
|
|
# machines in invented networks, which is the point of the program. The same
|
|
# reason the dashboard excuses `modules/dnsfilter`, where the feature itself
|
|
# is about ranges.
|
|
#
|
|
# This is deliberately broad. A narrower list would have to name every
|
|
# scenario file, and a new scenario would then have to remember this file —
|
|
# which is how an allow list becomes a list nobody keeps correct.
|
|
packages::internal-address
|
|
tools::internal-address
|
|
corpus::internal-address
|
|
CHANGES.md::internal-address
|
|
README.md::internal-address
|