Skip to content

Say PoE on the run that carries it - #83

Merged
sasagar merged 2 commits into
mainfrom
power-on-the-same-lead
Aug 4, 2026
Merged

Say PoE on the run that carries it#83
sasagar merged 2 commits into
mainfrom
power-on-the-same-lead

Conversation

@sasagar

@sasagar sasagar commented Aug 4, 2026

Copy link
Copy Markdown
Member

いったんは描けることが大事だと思うのでpoeを示せるようにしましょう

sw.LAN1 <-> cam.LAN : lan 30m "N-01" [poe]
図の札      N-01  30m  PoE
ケーブル表  | 番号 | 送出 | 受け | 信号 | 長さ | 給電 |
            | N-01 | PoE スイッチ | PoE カメラ | LAN | 30m | あり |
            | N-02 | PoE スイッチ | AP        | LAN | 15m | —    |

What was wrong with the two ways it could be written

Two linksduplicate-connection, and two rows on the cable schedule where one lead exists. Somebody loads the van from that.

poe over lan — clean, one row, but the run is named PoE and the network disappears from the drawing and the legend.

Neither says what is true: one Cat lead, carrying both. So PoE is a property of the run, not a run of its own.

What it buys

The questions that get asked on site: which ports have to be PoE, and which boxes need no power supply. Neither was readable before.

  • A show with no PoE never grows the column.
  • A run that is not Cat is reported — nothing puts power down a coax.
  • The carrier decides it, as everywhere else, so ndi over lan [poe] is fine.

Two things fell out

An attribute with no value is a flag. [poe] means [poe=true]. There is nothing to put on the other side of an = that reads better than saying it once.

The column needed a word of its own. A true printed as 未宣言 — that word belongs to a device nobody declared, and it says the opposite here. Columns now carry what their true means; the old one keeps the old default.

Named rather than hidden

The poe signal type is untouched, so nothing already written stops working. Having both a type and a flag for one thing is a seam. The changeset says so rather than leaving it for somebody to find.

392 core tests, seven of them new; full CI locally.

🤖 Generated with Claude Code

https://claude.ai/code/session_01824X53UWBk8JUvTPoN7hnu

sasagar and others added 2 commits August 4, 2026 22:21
Drawn as two links, PoE put two rows on the cable schedule where one
lead exists, with a duplicate-connection warning on top. Somebody loads
the van from that. Written as `poe over lan` it came out clean but named
the run PoE, and the network vanished from the drawing and the legend.

Neither says what is true: one Cat lead, carrying both.

    sw.LAN1 <-> cam.LAN : lan 30m "N-01" [poe]

One run, one row. The drawing shows PoE beside the number and the
length; the schedule gains a 給電 column, and a show with no PoE in it
never grows one. It answers what gets asked on site — which ports have
to be PoE, and which boxes need no power supply.

A run that is not Cat is reported; nothing puts power down a coax. The
carrier decides it as everywhere else, so `ndi over lan [poe]` is fine.

An attribute with no value is a flag now. There is nothing to put on the
other side of an `=` that reads better than saying it once.

The column needed a word of its own: a `true` used to print as
"undeclared", which belongs to a device nobody declared and says the
opposite here.

The `poe` signal type is untouched, so nothing already written stops
working. Having both a type and a flag for one thing is a seam, and the
changeset says so rather than leaving it for somebody to find.

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01824X53UWBk8JUvTPoN7hnu
@sasagar
sasagar merged commit d6ce614 into main Aug 4, 2026
1 check passed
@sasagar
sasagar deleted the power-on-the-same-lead branch August 4, 2026 22:35
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant