Cardano Updates

Signed-off-by: Aaravanand00 <[email protected]>

Signed-off-by: Pat Losoponkul <[email protected]>

* Added cardano-tracer-0.4.0

From https://github.com/IntersectMBO/cardano-node at b2b5692af3e8c699e9f70ec9789f2c46007c8a5a

* Added cardano-timeseries-io-1.0.0

From https://github.com/IntersectMBO/cardano-node at b2b5692af3e8c699e9f70ec9789f2c46007c8a5a

Signed-off-by: Hyperledger Bot <[email protected]>
Co-authored-by: Hyperledger Bot <[email protected]>
Co-authored-by: Pat Losoponkul <[email protected]>

Replace mkUnsignedTx + mkByronWitness/mkShelleyWitness with
mkLedgerTx + mkByronWitnessLedger/mkShelleyWitnessLedger in
the duplicated test module.

All 8 golden hex string tests pass — ledger-native construction
produces byte-identical CBOR to the cardano-api construction.

The builtin availability check in `scriptCBORDecoder` was using
`Set DefaultFun` for O(log n) membership, with a TODO suggesting
`IntSet`. A `UArray DefaultFun Bool` is a better fit: since
`DefaultFun` derives `Ix`, lookup is a true O(1) unboxed array index
with no conversion at the lookup site.

The array is built once per call via `runSTUArray`, folding over the
`Set` through its `Foldable` instance (no intermediate list). The
array covers all ~100 `DefaultFun` constructors.

Picks up the rescan-for-late-subdirs fix. Built by publish-images
CI from the preceding commit, which introduced it.

main() was snapshotting /opt/cardano-tracer/logs once at startup and
spawning one watcher per subdir present at that instant. Node subdirs
created *after* that snapshot (typical for producers that handshake
with the tracer slightly after the first relay, or for any node
restarted mid-run by Antithesis fault injection) were never tailed,
so their LogMessages never reached the Spec rules in Sidecar.hs. That
is why 'Sometimes: Any p{1,2,3} log' tripped on every run even though
the tracer was clearly writing those subdirectories to disk
(log-tailer, which polls independently, sees them fine).

Replace the one-shot listDirectory + forM with a forever loop that
maintains a Set of seen subdirs, spawns a watcher for each newly-seen
entry, and logs 'Tailing node log dir: <path>' as it does so. The
SDK 'reached' event for 'find log files' fires once, on the first
non-empty diff, preserving existing semantics for that assertion.

Verified locally: starting the sidecar with an empty logs dir and
then creating relay2, p1, p2, p3 subdirs one by one spawns a watcher
per subdir in order.

Fixes #50

`makeSizedByteStrings` used the same `H.Seed` for every element, so each
generated ByteString was a prefix of the same deterministic byte sequence.
Use `unfoldr (Just . Seed.split)` to produce a stream of independent
SplitMix seeds instead, giving uncorrelated content across sizes.

Picks up the rescan-for-late-subdirs fix. Built by publish-images
CI from the preceding commit, which introduced it.

main() was snapshotting /opt/cardano-tracer/logs once at startup and
spawning one watcher per subdir present at that instant. Node subdirs
created *after* that snapshot (typical for producers that handshake
with the tracer slightly after the first relay, or for any node
restarted mid-run by Antithesis fault injection) were never tailed,
so their LogMessages never reached the Spec rules in Sidecar.hs. That
is why 'Sometimes: Any p{1,2,3} log' tripped on every run even though
the tracer was clearly writing those subdirectories to disk
(log-tailer, which polls independently, sees them fine).

Replace the one-shot listDirectory + forM with a forever loop that
maintains a Set of seen subdirs, spawns a watcher for each newly-seen
entry, and logs 'Tailing node log dir: <path>' as it does so. The
SDK 'reached' event for 'find log files' fires once, on the first
non-empty diff, preserving existing semantics for that assertion.

Verified locally: starting the sidecar with an empty logs dir and
then creating relay2, p1, p2, p3 subdirs one by one spawns a watcher
per subdir in order.

Fixes #50

- Add <link rel="preload"> for both Chivo TTFs in headTags so the browser fetches them in parallel with HTML/CSS instead of waiting for the stylesheet
- Move @font-face declarations from src/css/custom.css to inline <style> in headTags so the URLs stay as /fonts/* and match the preload (otherwise webpack hashes them and the preload misses)
- Fix typo in insights/supply canonical URL: was https://www.cardano.org.com/insights/supply, now https://cardano.org/insights/supply/

- Add Required Frontmatter section to docs/get-involved/create-a-news-article with field table, good and bad examples, and clarification that description is distinct from the in-post body summary
- Add description to all four frontmatter examples in the doc
- Backfill description on the 12 most recent news posts (Mar 18 to Apr 17 2026), keeping each under 160 characters with concrete keywords for Google snippet and OG/Twitter cards
- Replaces the previous Docusaurus fallback that used the full post body, which got truncated mid-sentence in search results

- Add ESLint 10 flat config with eslint-plugin-react, react-hooks, jsx-a11y
- yarn lint and yarn lint:fix scripts
- Fix 24 jsx-a11y errors: 2 iframe titles, FAQ refactor (use Collapsible callbacks), modal backdrops with documented Escape fallback, 4 GovernanceCharts trigger areas with role/tabIndex/onKeyDown, 4 insights slider drag zones marked aria-hidden
- Fix 3 latent bugs: undefined ToolsTags reference, var projects redeclare, error rethrow without cause
- Disable noisy stylistic rules (no-unused-vars etc.) for now; cleanup deferred to a follow-up branch
- Real react-hooks antipatterns remain as warnings (75 total), to address in focused PRs later

Co-authored-by: github-actions[bot] <github-actions[bot]@users.noreply.github.com>

This allows for a better comparison and stresses that linear leios does
not result in more costs when load is < Praos capacity.