tickets

5.1 KB

1. # Report 010 (w080): tickets code order — no behaviour change
3. **Where:** commits on main in loreana:/media/STORAGE/projects/tickets.worldapi.org:
4. - `97e269b`: file moves
5. - `e9d5c61`: topics, map and thin wrappers
6. - `a75e027`: `let`
7. - the docs + tests commit, which also holds this report
9. Not pushed, not deployed. No `.env` read. Only ports 8750–8759 were used. Every server was started by a script that stops it again by its PID; nothing of mine is still running.
11. ## Counts (before → after; `python3 tests/letcount.py .`)
13. | | before (38f9d10) | after |
14. |---|---|---|
15. | `.hl` files in the root | 14 | **1** (project.hl) |
16. | project.hl lines | 621 | **144** (index comment, config, routes, audience) |
17. | `let` total | 680 | 219 |
18. | `let` never reassigned | **452** | **0** |
20. The 219 `let`s left are all needed:
21. - 136 are reassigned;
22. - 83 are re-bound in a loop body. Hybriel refuses a plain declaration there on the 2nd pass, which I tested.
24. ## Layout
26. The full file map is in README "Files". Imports only go downwards: util ← users ← projects ← tickets-helpers ← events ← tickets ← invites / connections ← api-helpers ← api ← project.hl.
28. - **`lib/tickets.hl`**:
29. - tickets and their relations, including every write to the tickets and links tables;
30. - `removeMember`, which also unassigns the person's tickets;
31. - `ticketRights`, moved out of the ticket page's server handler (face).
32. - **`lib/events.hl`**: the history, plus `putEvent`.
33. - **`lib/projects.hl`**: projects, members and roles.
34. - **`lib/tickets-helpers.hl`**: state names, the shapes a page shows, the filter links.
35. - **`lib/users.hl`**, **`lib/invites.hl`**: `invites.hl` gained `mayInvite` and `joinByInvite`, taken out of project.hl and the settings face.
36. - **`lib/connections.hl`**, **`lib/markdown.hl`**, **`lib/mdview.hl`**, **`lib/migrate.hl`**.
37. - **`lib/import.hl`**: was `ticketfile.hl`.
38. - **`lib/util.hl`**:
39. - was `localtime.hl`;
40. - also holds env, the storage folder, URLs, `countOf` / `first` / `merged`, the insertion sorts and the text checks;
41. - before, these were in both store.hl and users.hl.
42. - **`lib/api.hl`**: every function route, as thin wrappers.
43. - **`lib/api-helpers.hl`**: the old api.hl plus auth, filters, the Markdown `Accept` check and the HTML page.
44. - **`lib/jsoncheck.hl`** stays a file of its own: it is the workaround for hybriel #12 and gets deleted once that is fixed.
45. - **Moved:** `tools/import.hl` and `components/styles.hl`. The gate now calls `tools/import.hl`.
47. ## Tests
49. | test | result |
50. |---|---|
51. | gate `tests/browser.mjs` (8750–8759) | **249 passed, 0 failed** after each step |
52. | `tests/connect.mjs` (temp copy, app URL :8752, Chrome 8753–8759) | **60 passed, 0 failed** after each step |
53. | live-data reads | **0 different** |
54. | live-data writes | **0 different** |
56. **Live-data comparison** (new `tests/realdata-baseline.mjs` and `tests/realdata-compare*.py`; how to run: README "Test" → "Same output"):
57. - **Data:** a copy of Byrodin's storage taken 2026-10-03. Each run gets a fresh copy.
58. - **What it fetches:**
59. - every page twice: signed in as az5b2 through a session file, and signed out;
60. - every API read, as JSON and as Markdown;
61. - the refusals;
62. - then 72 writes (API calls and server handlers) and the reads that follow them.
63. - **Reads:** 2,172 answers.
64. - 1,460 are byte-identical.
65. - 712 differ only by source positions (`site`, mount keys like `Markdown@135:54`), the `?v=` hashes and the seed time.
66. - **Writes:** identical once the new ids and times are masked.
67. - **The masks change nothing on their own:** two runs of the OLD code compare clean with the same masks.
69. **Other checks:**
70. - A server started without `TICKETS_STORAGE` still reads the app's own `storage/mpackdb` (11 projects).
71. - `tools/fix-import-summaries.hl` gives the same output as the old tree on a copy.
72. - `tools/migrate-short-ids.hl` loads and refuses as before when it has no key.
73. - `tests/markdown.hl` gives the same output as before.
75. ## Found (Hybriel behaviour; written into STATUS "Lessons")
77. 1. **A `let` in a loop body cannot simply become a plain declaration.** Hybriel refuses it on the second pass.
78. 2. **Login sessions were lost after the first split.** The gate failed with 33 passed / 17 failed: every login through `/login/callback` lost its session.
79. - **Cause:** the request object holds the session (an instance). It survives one call. Copied into a second call (route wrapper → `loginCallback`), the instance was copied too.
80. - **Fix:** project.hl passes `&req` and `&server.sessions`. After that the gate was green.
81. - Reproduced in `.scratch/w080/reft/main4.hl`.
82. 3. **Import cycles are refused**, so the lib files import in one direction only.
84. ## Not done / open
85. - Not deployed (that is the architect's job). To deploy, use the normal `deploy.sh`. There is no data change.
86. - `tests/browser-pre20-parked.txt` and `tests/short-id-switch.mjs` were not run (both are historical). `short-id-switch.mjs` now points at `lib/users.hl`.
87. - Scratch (gitignored) in `.scratch/w080/`:
88. - the old tree (code only);
89. - `gate.sh` / `connect.sh` (the commands for ports 8750–8759);
90. - `unlet.py`, the one-off script that removed the `let`s.
91. - The live-data copy and every baseline output (they hold live data) were deleted after the run.