Skip to content

docs(landing): bring the docs site to 0.10 parity - #16

Merged
enekos merged 1 commit into
masterfrom
docs-landing-parity-0.10
Sep 1, 2026
Merged

enekos merged 1 commit into
masterfrom
docs-landing-parity-0.10

Conversation

@enekos

@enekos enekos commented Sep 1, 2026

Copy link
Copy Markdown
Owner

The docs page had drifted several releases behind the framework. This is a docs-only change — landing/src/Docs.svelte and landing/src/Landing.svelte, nothing under crates/.

Things that were wrong, not just missing

Claim on the site Reality
ai is a default feature Removed in 0.6.0; the tool surface is always-on core
Queue is "Postgres-backed, cross-pod" Runs on any Backend since 0.9.0
Queue::new(pool), handler takes |payload| Queue::new(store), handler takes &JobCtx
Pg::from_env() Takes a pool size: Pg::from_env(8)
Migrations are hand-written closures The diff/generate/drift workflow was entirely absent
362 KB core binary 394 KB measured

Pillars that had no documentation at all

WebSockets · PubSub · channels & presence · actors & supervision · the REPL · relations · JSON path queries · full-text & hybrid search · embeddings & vector search · reactive queries · backend capabilities · row locks / isolation / RETURNING / bulk insert · advisory locks · collections · crypto · static files · credentialed CORS · ops_guard · App::listener

Internals & options

These existed only in the source. Added: the request lifecycle and guard ordering, why keep-alive is capped twice, the two slowloris deadlines, panic isolation and its panic = "abort" caveat, how an upgrade detaches from the worker, where ~394 KB comes from — plus tables for every Limits and WsConfig field with its default, the per-backend capability matrix, the feature matrix, and the 22-crate workspace map.

53 sections in 7 groups, up from 30 in 6.

Verification

  • pnpm run build clean; svelte-check 0 errors / 0 warnings.
  • Headless render: 53 nav links against 53 article anchors, 0 broken deep-links; no horizontal page overflow (scrollWidth == clientWidth); all six escaped-brace literals ({{ escaped }}, {a,b,0}, …) render as literal braces.
  • pnpm run check still exits non-zero, but on a pre-existing vite.config.ts error — Vite 8 dropped brotliSize. Untouched here; worth a separate one-line fix.

Note for the reviewer

The "Responses & errors" section documents 5xx bodies being redacted to {"error":"internal error"}. That behaviour is in e2a6fd8, which is still unpushed on local master — this section becomes accurate the moment that commit lands.

Out of scope, still stale: README.md lists 11 of 22 crates, and its Status "Next:" line names keep-alive, form-encoded bodies and relations, all of which have shipped.

The docs page had drifted several releases behind the framework. It still
advertised an `ai` feature (removed in 0.6.0), described the queue as
Postgres-only with the pre-0.9 `Queue::new(pool)` / `|payload|` API, showed
`Pg::from_env()` without its pool argument, and documented migrations as
hand-written closures only — no sign of the diff/generate/drift workflow.

Whole pillars had no page at all: WebSockets, PubSub, channels and presence,
actors and supervision, the REPL, relations, JSON path queries, full-text and
hybrid search, embeddings, reactive queries, backend capabilities, row
locks/isolation/RETURNING/bulk insert, advisory locks, collections, crypto,
static files, credentialed CORS, ops_guard, and App::listener.

Adds the internals and options the framework never explained anywhere but its
source: the request lifecycle and guard ordering, why keep-alive is capped
twice, the two slowloris deadlines, panic isolation and its panic=abort
caveat, how an upgrade detaches from the worker, plus every Limits and
WsConfig field with its default, the per-backend capability matrix, and the
crate map.

53 sections in 7 groups, up from 30 in 6. Landing page: fixed the 362 KB
figure (394 KB measured), the Cargo snippet, the queue feature card, the
stale jobs tutorial step, and added a realtime step.

Verified: vite build clean, svelte-check 0 errors, and a headless render
confirms 53 nav links against 53 article anchors with no broken deep-links,
no horizontal overflow, and every escaped brace rendering literally.
@enekos
enekos marked this pull request as ready for review September 1, 2026 18:04
@enekos
enekos merged commit fa4c954 into master Sep 1, 2026
1 of 2 checks passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant