# DBSC Summer Series 2026 Dublin Bay Sailing Club's 2026 course cards for keelboats racing on fixed marks, as published at [dbsc.ie/racing/2026-course-cards](https://dbsc.ie/racing/2026-course-cards/): | File | Source | Made by | |---|---|---| | `marks.json` | `source/DBSC_Marks_Bearings_Distances_2026_v1.pdf` (+ three marks from `source/machine-readable/DBSC_Marks_2026.csv`) | `tools/extract_dbsc_marks.py` | | `cc1-saturday-cv.json` | `source/CC1_Saturday_CV_2026_v1.pdf` — Saturday, committee vessel starts | `tools/extract_dbsc_card.py` | | `cc2-saturday-hut.json` | `source/CC2_Saturday_Hut_2026_v1.pdf` — Saturday, West Pier hut starts | `tools/extract_dbsc_card.py` | | `cc3-thursday-blue.json` | `source/CC3_Thursday_Blue_2026_v1.pdf` — Thursday, Blue Fleet | `tools/extract_dbsc_card.py` | | `cc4-thursday-red.json` | `source/CC4_Thursday_Red_2026_v1.pdf` — Thursday, Red Fleet | `tools/extract_dbsc_card.py` | | `cc5-tuesday-hut.json` | `source/CC5_Tuesday_Hut_2026_v1.pdf` — Tuesday, hut starts | `tools/extract_dbsc_card.py` | | each card's `startLine` | `source/SI-B_…_Committee_Vessel_2026_v1.pdf` (cards 1, 3, 4) or `source/SI-H_…_Hut_2026_v2.pdf` (cards 2, 5) | `tools/extract_start_line.py` | | `source/*.md` | the sailing instructions below | `tools/pdf_markdown.py` | | `*.html`, `map/marks.svg` | the JSON above, `map/background.png` | `tools/render-cards.ts` | | `map/background.png`, `.json` | OpenStreetMap + OpenSeaMap tiles | `tools/fetch_map.py` | All the club's course-card documents are version 1, dated 7 April 2026 (the Red Fleet card 4 April); the sailing instructions kept alongside them carry their own versions and dates, listed below. `source/` keeps them verbatim; `source/machine-readable/` is the club's "Machine-Readable Course Cards and Marks" zip unpacked — a CSV per card, a marks CSV and GPX, and the club's `Warning.txt`: *"These files are not official DBSC documents. They have not been tested thoroughly … If there is any difference between any of these files and the corresponding official pdf document, the pdf document should be taken as correct."* The JSON is therefore generated from the PDFs, and the machine-readable files are used to check it (below). `source/CC-DBSC-Racing-Marks-Chart-2025.pdf` is the club's chart of the marks, kept for reference: the club notes it has not been revised for 2026 and shows approximate positions only. `manifest.json` records each artifact's source, the URL it was fetched from, the metadata that heads the output, and the cross-checks; `pnpm data` rebuilds everything and `pnpm data:check` verifies the committed files against a fresh run — the cross-checks run in both. ## How the JSON is produced All the PDFs are real text, so extraction is a matter of reading `pdftotext -bbox` word positions; nothing is OCR'd. **Marks.** The "Marks, Bearings and Distances" sheet has a row per mark, printed on two lines: name and latitude minutes, then colour and longitude minutes, with the letter between. The header says what the minutes are relative to ("Lat N 53°+", "Long W 6°+"); longitudes are West, so negative. Colours are kept as printed, lowercased ("yellow/black", "yellow nav mk" for the Turning mark, which is a navigation buoy); the sheet gives no shapes. Three marks the club lists are not rows of the sheet — the black and green start marks at the West Pier hut (`2`, `3`), which course card 3 uses, and Zebra (`Z`), which the chart notes "shows position only and may not be laid" and no 2026 course uses — and are appended from the club's marks CSV. The sheet does have a column for `3` (bearings and distances *to* it), which is checked against the CSV position below. **Cards.** Each card is 16 sections lettered A–R (no I, no O), one per compass point, each headed with a bearing (`A 000°`, `B 022°`, …) and listing courses 1–8 (1–5 on the Tuesday card). A course is a row of marks, each a letter or digit suffixed `p` (printed red, rounded to port) or `s` (green, starboard). The extractor finds the section headings (a `ddd°` word with a capital letter to its left), takes their x positions as the columns, reads each numbered row within its column, and refuses any word in a row that is not a mark. The course id is letter + number (`B3`); the format stores the id as printed and leaves the club's convention to the club, and the section headings are carried as the card's first note, "Course sections". The Red Fleet card prints bare letters and says "All marks to be rounded to Port", which the extractor applies; on the two hut cards the notes say the Turning mark X is "to be passed to Port" rather than rounded, so it is a passing mark there — on card 3, which also uses X, the notes say nothing special and it is a rounding mark like the rest. **Start lines.** The cards do not say where a race starts; the sailing instructions supplements do, and they differ by card. Supplement B 4.2 — cards 1, 3 and 4 — puts the line between a red and white staff on the committee vessel and a starting mark or an orange-flagged RIB: laid on the day, so a `placement` and no position. Supplement H 4.1 and 4.2 — cards 2 and 5 — put it at the West Pier hut, on "a transit formed by bringing in line the two triangles above the West Pier Hut", crossed between the hut and whichever limit mark is signalled: a fixed line, though the club publishes no coordinate for it, so it too is carried as the SI's words. Each card carries its own as a `startLine` and every course begins there, which is why the start line belongs to the card and not to the marks file the five share. `extract_start_line.py` refuses any text that is not in the SI verbatim. **Notes.** Every card carries its section headings as "Course sections", its own printed text — the lines under and beside the courses, less the title block — as "Card notes", and the marks sheet's caveat ("Mark positions may vary slightly. All figures are approximate. Warning: Some direct paths are obstructed.") after it. The Red Fleet card's "W/L courses 1-5 / RTC Courses 6-8" (windward–leeward, round the cans) is in its notes. ## The sailing instructions DBSC publishes its instructions as a general set and a supplement per fleet and course type, plus amendments through the season. The five kept here are the ones the keelboat fixed-mark cards depend on; each has a Markdown sidecar written by `tools/pdf_markdown.py`. All are at [dbsc.ie/racing/sailing-instructions-2026](https://dbsc.ie/racing/sailing-instructions-2026/). | File | What it decides for these cards | |---|---| | `SI-A_Sailing_Instructions_General_2026_v2.pdf` | A4.3: flag X ashore on a Saturday sends every keelboat class but the Green fleet to the hut, under supplement H and Course Card 2. A10.4 restricts passing between the hut and the start limit marks. | | `SI-B_Sailing_Instructions_Fixed_Marks_Committee_Vessel_2026_v1.pdf` | B4.2 the start line of cards 1, 3 and 4; B5.5 which card on which day; B5.6 the red/green "p"/"s" sides the cards print; B6 the finish. | | `SI-H_Sailing_Instructions_Fixed_Marks_Hut_2026_v2.pdf` | H4.1–4.3 the start line of cards 2 and 5 and its limit mark; H5.5 which card on which day; H5.6 Turning mark X passed to port, not rounded — which is why X is a passing mark on those two cards and not on card 3; H6.1 the finish, also at the hut. | | `Sailing_Instructions_2026_Amendment2_CV_starts.pdf` | Adds B4.3: if the committee vessel cannot anchor, the line may be between two marks, or a mark and a RIB. An alternative to B4.2, not a replacement, so the cards' start line quotes B4.2. | | `Sailing_Instructions_2026_Amendment4.pdf` | Adds B5.12: on Saturdays 29 August and 12 September the Red Fleet and Dublin Bay 21s race Course Card 4 from Corinthian in Scotsman's Bay, and a course may be repeated (`2A3`) or two combined (`A3B8`). The JSON holds the courses as printed and leaves combining them to the reader, as it does the club's other signalling conventions. | The club's other instructions are not kept: supplements C (coastal races), D (dinghies) and G (Green fleet) and the Water Wag and Greystones feeder instructions cover fleets with no card here, and amendments 1 (large-ship priority) and 3 (dinghy starts) change nothing a card depends on. ## How it was checked `tools/check_dbsc.py` runs as part of `pnpm data` and `pnpm data:check`: - **Cards against the club's CSVs.** Every course on all five cards — 592 courses — agrees with the machine-readable CSV in marks and sides, with **one exception: course K1 on card 2** (Saturday, hut). The PDF prints `Fs Gs Vs Cs Es Bp Gs Vs Hp`; the CSV ends it `… Hp Xp`, with the Turning mark that every other course on that card finishes with. The PDF cell is not clipped — the nine marks are spaced out to fill it — so the card as published omits X, whether by design or not. The club says the PDF is correct, so the JSON follows it; the difference is recorded as `expected` in the manifest, and the check will fail if either document changes. - **Marks against the CSV and GPX.** All 26 marks: same ids, names and positions, to the CSV's own rounding. - **Positions against the sheet's bearings and distances.** From the extracted positions, the true bearing and distance between every pair of marks on the sheet — 529 pairs, including the `3` column — reproduce the printed figures within 1° and 0.01 NM, i.e. to their printed precision. So the sheet's positions, its table, the CSV and the GPX are all one consistent set (the sheet says its bearings are true, and they are). - Structural tests confirm each card's 16 × 8 (or 16 × 5) courses in the card's order, every mark on the marks file with a side, the Red Fleet card all to port, X passing and last on the hut cards (K1 excepted), that every course begins at the card's start line and that the line quotes the right supplement, and spot checks of a dozen courses read from the printed cards by eye. ## Notes on the cards - Start lines: the committee-vessel cards (1, 3, 4) start at a committee vessel whose position is a race-day fact; the hut cards (2, 5) start on a transit from the West Pier hut, crossed between the hut and the limit mark signalled — one of the orange, black or green marks (`O`, `2`, `3`) or Turning mark `X` under H4.3; the cards' own notes add Omega `Y`. Neither line is printed on a card, and the club publishes no position for the hut, so both are carried as the SI's words and placed per race. - The club's page notes for 2026: Merrion (`L`) has shifted slightly for marine survey work, Battery's (`T`) position has been updated, and Omega (`Y`) will not be a data buoy this year. - The section bearing is the wind direction the section's courses are set for — A north, B 022½°, … R 337½°, printed rounded down — so a race officer picks the section by the wind and a course by length. The JSON carries the headings as printed, in the "Course sections" note, not as a property of the course: what the wind is doing on the day is the caller's knowledge, not the card's.