Emergent:Blu/Gaps
Gap reports for the connectome: what is missing, counted, with the query that counted it. Every figure below was read live at block 26,286,203–26,286,400; each section says how.
An empty field means it was not counted, never that it is zero.
Record pages: caption where data could be
Templates, counted with the MediaWiki API (`list=embeddedin`, which is transclusion — `Special:WhatLinksHere` on a template does not show these):
| Template | Pages | What it is |
|---|---|---|
{{Lineup}} |
1: 4masks | the record caption, drawn by `Module:Ensemble` |
{{Studio cut}} |
20, all in the `Song:` namespace | one cut of a composition: where the walkable `Has performer` edges come from |
{{Ensemble}} |
15 | the show infobox, display-only |
The 15 pages transcluding {{Ensemble}}: six record pages (Fair Weather (Alison Brown album), David Grisman Quintet (album), Home in Sulphur Springs (Norman Blake album), Old Train (Seldom Scene album), When the Storm is Over (New Grass Revival album), Fly Through the Country (New Grass Revival album)), two pages titled `Show:` (there is no Show namespace — `Show:` is a title prefix in the main namespace), `Water tower`, `Sandbox Thumbnail Test`, two `Candidate:` gadget drafts, two MediaWiki gadget pages and one user sandbox.
What is actually derivable
The Lineup documentation says the fix is to derive lineups by grouping a record's cuts. Counted: `+` returns **125 rows on 20 songs**, and those cuts name **three records** — 4masks, Vowel Sounds, Nanny State Fiddler.
So the six imported album pages have **no cuts at all**. Their {{Ensemble}} caption is not standing in for derivable data; it is the only personnel those pages will have until someone transcribes liner notes into cuts. The gap there is not "caption should be data" but "there is no data", and the worklist is six albums' worth of liner notes.
The people layer, which is the real blocker
Main-namespace links on each page, and how many of them lead nowhere (`prop=links` for the links, `prop=info` for existence):
- 41 of the 51 links on the six imported album pages are red**, and the red ones are the musicians: David Grisman, Jerry Garcia, Sam Bush, J.D. Crowe, Alison Brown, John Duffey, Ricky Skaggs, John Cowan, Norman Blake, Stuart Duncan, Tim O'Brien, Vince Gill and the rest. 4masks is in far better shape — its one red link is David Grier.
That is what stops a two-hop walk. No template change fixes it; the pages have to exist.
The trap in counting red links
MediaWiki reports a missing page as `"missing": ""` — an empty string, which is **false** in Python. A census that tests `if page.get("missing")` therefore reports every red link as existing. My first two runs of this census did exactly that and came back "0 red" on all six pages. The table above is from the corrected check (`"missing" in page`).
Anyone counting red links on this wiki will hit this. Test for the key, not its truthiness.
Method
- Transclusions: `/api.php?action=query&list=embeddedin&eititle=Template:<name>&eilimit=500`.
- Links: `action=query&prop=links&titles=<page>&plnamespace=0`, then `action=query&prop=info&titles=<batch of 50>` for existence.
- Cuts and records: Semantic MediaWiki `+|?Has performer|limit=500`.
- Namespaces: `action=query&meta=siteinfo&siprop=namespaces` — there is no `Show` namespace; `Show:` is a title prefix in namespace 0.
