Skip to content

Review: chapter 13 Lessons learned (step D.5)

A source

This page is a source of edakit – a concept, a prompt or the rules of a session –, shown as it is in the repository. Read it as such; the guides say what edakit does.

Status: built 2026-11-01 – chapter 13 is the history of edakit and holds what the other chapters moved out; for the owner's review.

What changed

  • Measures (measure.py against HEAD): lines 112 → 255; own words 1073 → 2636; mean sentence 21.9 → 24.6 words; "the reference project"/"the user" 5 → 0; sections, paragraphs, bare references, dates: 0 → 0.
  • First paragraph: for whom (anyone who wants to know where edakit comes from and why its rules are as they are) and what they find.
  • pytree described once, in the second paragraph, linked to docs/projects/pytree.md: a project of about 120 parts built in October 2026, with the board description that chapter 1 ("What it achieved") gave. Elsewhere "pytree" or "the project".
  • Words: "the reference project" → "pytree"; "the user" → "someone" (13.2 items 2 and 8) and "the person who decided" (13.3, numbered decisions); "the owner's accounts/decision" → "our accounts", "Decided:" (the chapter speaks as "we").
  • Headings: 13.1 to 13.6 unchanged. New: 13.7 After the first board, day by day (a dated table, 22 Oct to 1 Nov) and 13.8 How the chapters' details came about with ### per group of chapters.

Where each history file went

File Passage Now in
01 "All of these happened or nearly happened …" 13.8, chapters 1 and 3
01 "What it achieved": the board, five days 19–23 Oct, the redesign in a day, one order, no prototype round intro (board, one order), 13.5 (19 to 23 October), 13.1 (redesign, already there)
01 the list of findings, "Each became a check" 13.8 intro, merged with 14's "The rules of edakit are lessons"
02 nothing moved –
03 graph.json until October 2026; the cathode found by the name "K" 13.8, chapters 1 and 3
04 dates of PARTS.library decision 8, dc_bias sheets, the six wrong symbols, the live run of 3.6 13.7 (31 Oct rows)
04 "some 30 searches"; the LED footprint story; the build using the bell's and pytree's calls 13.8, chapters 4, 6 and 7
05 three "as before" remnants not taken – the file says they need no place (no fact of edakit's past)
06 template named after its page (LT-10, 2026-10-31) 13.7 (31 Oct)
06 empty netlist saved; copySchematic copying the bell's ESP32 THT template 13.8, chapters 4, 6 and 7
06 timing 123 parts: placing 2.1 min, connecting 3.7 min 13.2, step times (merged with the 6 min already there)
06 pin on two nets, "regenerating fixed it" 13.2 item 6 (merged)
07 slide switch hour; geometry among the artifacts until October 2026; 70 of 178 open with the ground plane; 320 false DRC errors; Import Changes holding the gate in ten runs, once 2¾ h 13.8, chapters 4, 6 and 7
07 autorouter 1773 → 223 lines, a day of repairs converging nowhere 13.4, "Freerouting from day one" (merged)
07 one GND link open three times after a part was recreated 13.3, "Freerouting with our own DSN" (merged)
08 M-5 lesson; GEO.pour (45 % bare); looks row measured by hand; one-file edakit.toml until October 2026 13.8, chapter 8
08 footprint geometry among the artifacts until October 2026 merged with 07's into one item, 13.8
08 decisions 2026-10-30 (nothing orders or pays; fee table, step 5), 2026-10-31 (no regulatory content; [looks] mark), 2026-11-01 (wishes to makers); stage E since 2026-10-31; the state line of 8.4 (step 0, cart live at four makers) 13.7
08 tooling holes found only in the quote 13.4, merged into the existing item
08 four boards published as a draft by a script of its own 13.6, as the bell's (see Assumptions)
09 opening paragraph (3.2.206, 1.4.0, 19–26 Oct; 3.2.209 on 1 Nov) 13.8, chapter 9
09 the dates out of the tables 13.7, one row each (22 Oct, 26 Oct, 30 Oct, 31 Oct, 1 Nov)
10 decision of 2026-10-31 on mark 13.7, merged with 08's same decision
11 "a week of daily use"; elektrotechniker/pcb-pruefer; coordination tool; agreements reach every session 13.8, chapter 11
11 parallel sessions normal; 20-minute run lost; 38 and 93 minutes waiting; 49 minutes for a click 13.2 items 1 and 2 (merged; the 93 minutes are the table's connector swap)
11 the project went on growing after the extraction (0 Ω bridges, star LED, pour-only nets) 13.4, "Generic from the start"
12 nothing moved –
14 pcb-pruefer reviews rounds and signs M-5 13.8, chapter 11 (merged with 11.1's)
14 REQ.missing and REQ.missing:cc from the buzzer stage and USB-C input 13.8, chapters 14, 16 and 17 (the CircuitSmith half stays in 14.6)
15 "as before" dropped not taken (no fact)
16 the first parity run (/GND); two things learnt on route all 13.8, chapters 14, 16 and 17
17 the rename into codes and the re-sign-offs of the same day 13.8, chapters 14, 16 and 17
18 nothing moved –

Merged

  • The 31 Oct live run of 3.6 (from 4.5) and "Live on 2026-10-31 (step 3.6): the pads come back as built" (from 9.4) are one row.
  • The [looks] mark decision (8.4 and 10.2) is one row with the electrical-hash decision of the same day.
  • The 1 Nov 3.2.209 rename sentence of 9.3 went into the row of the live check on 3.2.209.

Assumptions

  1. The four published boards are the bell's. Chapter 8 said "The reference project built this as a script of its own first and published its four boards – the main board and the variants smd, tht and esp32". main, smd, tht and esp32 are the bell's variants (docs/projects/christmas-bell.md), pytree has one tree board, and commit a0bfb90 names the bell's D-22. Chapter 13 puts it in 13.6 as the bell's. If it was pytree after all, move the item.
  2. The second paragraph's board comes from chapter 1's "What it achieved" (MCU module, three LED driver ICs, about 30 LEDs, buttons, DIP switch, buzzer, concave outline, assembled by JLCPCB on one side); pytree's page agrees (RP2350-Zero, three LED drivers, 30 LEDs, back side assembled).
  3. "We" for the owner. Decisions read "Decided:" and the maker accounts "our accounts"; the style guide keeps "the owner" off reader pages.
  4. "the user" → "someone"/"the person who decided" rather than "you": these are past events, not instructions to the reader.
  5. "found the hard way" → "found by trial" in the chapter 9 opening; "again the hard way" (11.5) → "learn them again by trial".
  6. 13.7's title "After the first board, day by day" although its first row (22 Oct, the DRC rule set) falls in pytree's week: the row is a date that stood in chapter 9, not part of 13.1's story. 13.1 is unchanged.
  7. Context sentences not taken: 17.md's "The note names R-2 …" and the generic statements of 8.2 that the history files carried as context only stay in their chapters; 13.8 takes the story part.
  8. The stories inside 11.4 stay there (11.md: moving them changes the agreements and is the owner's decision).
  9. Mean sentence 24.6 words: 13.2's step-time sentence (73 words, unchanged from before) and the facts of the slide switch keep it up; the sentence rule holds only for the entry pages.

Findings

  • tests/test_style.py RED lists handbook/13-lessons-learned.md under "words"; the chapter now meets the rule, so words:handbook/13-lessons-learned.md is XPASS(strict) and fails until it leaves RED (not changed here).
  • Chapter 8's sentence on the four published boards (see Assumption 1) named the wrong project, if the assumption holds.