- extend vue-seo-layer (link dependency during development) and drop
@nuxtjs/seo: the page head, canonical and the one JSON-LD graph come
from the CMS resolve answer; the post's own title, excerpt and image
apply only while there is no CMS answer
- remove the journal sitemap source and the copied shared purge route,
library and plugin (the SEO layer serves POST /api/__site__/cache/purge);
the deprecated POST /api/__journal__/cache/purge keeps working
- post title is h1 while the Journal API reports content_h1_to_h2 on
(or absent) and p.journal-post__title while off; journal.postTitleTag
('auto' | 'h1' | 'p') overrides it; body h1 headings are demoted to h2
while the setting is on (gateJournalHtml/sanitizeJournalHtml demoteH1)
- journal.breadcrumbs option and post-title CSS tokens
- smoke harness: count=, header=, noheader= and resolves lines; the
fixture API also answers the SEO resolve and site endpoints
BREAKING CHANGE: the layer no longer installs @nuxtjs/seo (no
nuxt-site-config, nuxt-schema-org, robots or sitemap module, no
/api/__sitemap__/journal source, no useSiteConfig()); apps take their
head, sitemaps and robots from the SEO plugin through vue-seo-layer.
nuxt-site-config copies every NUXT_SITE_* variable into the public site
config, so NUXT_SITE_CACHE_PURGE_TOKEN reached every page's payload as
cachePurgeToken (and NUXT_SITE_CACHE_ALLOW_PROXIED as cacheAllowProxied).
- The purge settings move to runtimeConfig cachePurge.token
(NUXT_CACHE_PURGE_TOKEN) and cachePurge.allowProxied
(NUXT_CACHE_PURGE_ALLOW_PROXIED), outside that prefix.
- The old siteCache.* names keep working for one release: a non-empty
new token wins, either allowProxied flag turns it on, and one
deprecation warning per process names the variables (never a value).
- The shared Nitro plugin server/plugins/site-cache-purge.ts strips
cachePurgeToken and cacheAllowProxied from nuxt-site-config's stack on
site-config:init, so a site on the old names leaks nothing either; it
does nothing without nuxt-site-config.
- The shared lib, route and plugin and tests/sitePurge.test.ts are
byte-identical in vue-journal-layer, vue-menu-layer and vue-faq-layer,
with three sha256 pins.
- README: new names, the deprecation, the stripping and the advice to
rotate a token ever set under the old name.
- runtimeConfig.siteCache.allowProxied (NUXT_SITE_CACHE_ALLOW_PROXIED, default
false) lets the shared purge route accept a request that came through a
TLS-terminating proxy; the bearer token is then the only gate
- shared lib, route and spec stay byte-identical across the three layers,
with updated sha256 pins and two new sitePurge specs
- version 0.5.0
- server/lib/siteCachePurge.ts and the route, byte for byte from
vue-faq-layer (sha256 pinned in tests/sitePurge.test.ts)
- runtimeConfig.siteCache.purgeToken (NUXT_SITE_CACHE_PURGE_TOKEN) and the
'/api/__site__/**': { cache: false } route rule
- /api/__journal__/cache/purge, app/utils/siteCache.ts and
journal.cachePurgeToken unchanged, documented as aliases for one release
- version 0.4.0
Nitro's cached handler renders from a header-less copy of the request, so
under swr route rules useJournalApi, /rss.xml and the sitemap source
forwarded only the Nuxt peer and every cache miss shared one Journal API
rate-limit bucket (deferred in Phase 16). A Nitro plugin records
x-forwarded-for plus the peer in event.context.journalForwardedFor on the
request hook; the three callers forward it. Same key as the host app's
workaround, which keeps working. The acme SSR smoke's x-forwarded-for check
on the now route-cached playground is the test.
- POST /api/__journal__/cache/purge removes every Nitro route-cache entry
- 404 without a 32-character token or behind a proxy, 401 on a wrong bearer
token (constant-time over sha256 digests)
- routeRules keep /api/__journal__/** out of the route cache
- runtimeConfig.journal.cachePurgeToken (NUXT_JOURNAL_CACHE_PURGE_TOKEN)
- fixture API reload(); testing/purge-check.mjs proves a cached page shows
a change after a purge; the playground caches with swr 300
- author pages resolve slug or login and canonicalise to the slug
- tag pages, per-page SEO meta and canonicals that keep ?page=N
- sitemap source pages the API and builds locs from journal.routes
- /rss.xml proxies the API feed with x-forwarded-for
- neutral README; version 0.1.0
