docs/providers/sqlite.md cites code by line number in 7 places. Nothing measures those numbers, so a citation is true only until the next insertion above it, and nothing goes red when it stops being true. A method name is greppable and survives an insertion, so it needs no correction round at all.
Measured 2026-09-06. Of the 5 citations that name a method beside the line, 5 point at the wrong line today:
getDatabasePath() cited :133, declared at :372
query() cited :159, declared at :490
validate() cited :67, declared at :348
The smallest of the remaining docs, and the one with the clearest internal contradiction: line 150 cites getDatabasePath() at sqlite.ts:133 and line 492 cites getCapabilities() at sqlite.ts:133. Two different methods cannot share a declaration line.
While you are here, docs/providers/libredb.md carries exactly one line citation (factory.ts:100, line 110). Fixing it in the same PR is welcome, and saves a one-line follow-up issue.
Fix. Rewrite each citation to name the declaration instead of its coordinate, keeping the link to the file:
`getSchema()` ([`sqlite.ts`](../../src/lib/db/providers/sql/sqlite.ts))
docs/providers/mongodb.md (#581), mssql.md, oracle.md and trino.md are already in this shape and are the model to copy. Then add the doc to the NAMED_CITATIONS list in tests/unit/provider-docs-monitoring-citations.test.ts, with src/lib/db/providers/sql/sqlite.ts as its source and the method names the doc now uses. That list's two tests then assert the doc carries no .ts:<line> anywhere and that every name it cites is really declared, so this cannot rot again. Read that file's header comment first: it explains the policy and why it is pinned rather than the coordinates.
Done when grep -oE '\.ts:[0-9]+' docs/providers/sqlite.md returns nothing, the doc appears in NAMED_CITATIONS, and the suite is green. One doc plus one list entry, no product code.
Tracked in docs/BACKLOG.md as DOC4.
Curated for Hacktoberfest 2026. Comment to claim it before you start so two people do not work on the same issue. A PR must reference this issue and land with its tests in the same change; see CONTRIBUTING.md. Repo rules that apply: run bun run test (never bare bun test), and the 100% line-coverage gate must stay green. Local bun run test also needs the helm binary on PATH for unrelated chart tests - see #570 if it bites.
docs/providers/sqlite.mdcites code by line number in 7 places. Nothing measures those numbers, so a citation is true only until the next insertion above it, and nothing goes red when it stops being true. A method name is greppable and survives an insertion, so it needs no correction round at all.Measured 2026-09-06. Of the 5 citations that name a method beside the line, 5 point at the wrong line today:
getDatabasePath()cited:133, declared at:372query()cited:159, declared at:490validate()cited:67, declared at:348The smallest of the remaining docs, and the one with the clearest internal contradiction: line 150 cites
getDatabasePath()atsqlite.ts:133and line 492 citesgetCapabilities()atsqlite.ts:133. Two different methods cannot share a declaration line.While you are here,
docs/providers/libredb.mdcarries exactly one line citation (factory.ts:100, line 110). Fixing it in the same PR is welcome, and saves a one-line follow-up issue.Fix. Rewrite each citation to name the declaration instead of its coordinate, keeping the link to the file:
docs/providers/mongodb.md(#581),mssql.md,oracle.mdandtrino.mdare already in this shape and are the model to copy. Then add the doc to theNAMED_CITATIONSlist intests/unit/provider-docs-monitoring-citations.test.ts, withsrc/lib/db/providers/sql/sqlite.tsas its source and the method names the doc now uses. That list's two tests then assert the doc carries no.ts:<line>anywhere and that every name it cites is really declared, so this cannot rot again. Read that file's header comment first: it explains the policy and why it is pinned rather than the coordinates.Done when
grep -oE '\.ts:[0-9]+' docs/providers/sqlite.mdreturns nothing, the doc appears inNAMED_CITATIONS, and the suite is green. One doc plus one list entry, no product code.Tracked in
docs/BACKLOG.mdas DOC4.Curated for Hacktoberfest 2026. Comment to claim it before you start so two people do not work on the same issue. A PR must reference this issue and land with its tests in the same change; see CONTRIBUTING.md. Repo rules that apply: run
bun run test(never barebun test), and the 100% line-coverage gate must stay green. Localbun run testalso needs thehelmbinary on PATH for unrelated chart tests - see #570 if it bites.