Release tooling and vocabulary release
Integrated nine new checks for release tooling, release evidence, tool checksums, heritage spreadsheets, Spanish search and vocabulary, the SKOS export and map labels into the 209-check production gate.
CHANGELOG ARCHIVE · PAGE 2 OF 7
Entries 51–100, newest first.
Integrated nine new checks for release tooling, release evidence, tool checksums, heritage spreadsheets, Spanish search and vocabulary, the SKOS export and map labels into the 209-check production gate.
Every release record since Phase 219 is now checked: a release is closed only with the owner's production evidence (deployment, READY, production and domains), and dataset releases must match it.
Map labels now sit above the dots, never overlap each other and keep strong contrast, fixing an accessibility issue where a label could fall under a neighbouring dot.
The smell descriptors (in English and Spanish) and the places by kind are published as SKOS concept schemes for museum, archive and library systems.
A new pilot page lists every descriptor in Spanish and English, with its review status and a Spanish search link.
Searching in Spanish for a descriptor such as ahumado or a guardado now finds the smells it describes, and phrasing such as olor a humo works too.
Institutions can keep olfactory heritage records in a spreadsheet template and check the filled file with the heritage record checker.
Each downloadable tool now has a published SHA-256 checksum, with instructions to verify it before running.
A helper checks production evidence before a release is closed, lists the tasks it closes and prepares the QA record.
A helper computes the dataset release entry and release notes for the exact merged commit, so knowledge-graph releases are recorded the same way every time.
Integrated nine new checks for the heritage checker, graph diff, scale rehearsal, decision log, place kinds, Spanish descriptors, standalone tools, contribution review and multilingual linked data into the 200-check production gate.
The linked-data graph now includes every descriptor, links smells to their descriptors, carries Spanish names as language-tagged literals with their review status, and gives each place its kind.
The contribution review model now runs as tested rules: no automatic promotion, no self-review, the right role for every decision and a registered source for verification. Contributions remain not live.
The heritage record checker and the graph export diff can be downloaded as single files and run with Node.js, without the SmellBook source code. They read only your files and send nothing anywhere.
All 37 descriptors now have Spanish names, pending native-speaker review, shown on descriptor pages and Spanish pilot pages.
Places are grouped into seven editorial kinds on /places, on each place page and in the map's place filter, and the graph export carries each place's kind.
The governance page now lists the owner decisions that shape SmellBook, with where each is recorded, and the decisions still open.
Identity, entry-standard and search code now run against 1,000 to 100,000 synthetic entries in memory. The duplicate-name check went from 63 seconds to 75 milliseconds at 10,000 entries, with identical results.
A tool compares two copies of the knowledge-graph export and lists the nodes and edges added, removed and changed.
Museums, archives and researchers can check draft olfactory heritage records, including every SmellBook identifier they link to, before sharing them.
Integrated six new checks for search filters, entity identifiers, backlinks, entry siblings, the downloads listing and sitemap alternates into the 191-check production gate.
Spanish pilot pages and their English versions now list their language alternates in the sitemap.
Every entry is now available as BibTeX and CSL-JSON for reference managers.
The downloads page now lists every public file with its type, size in records and a readable filename.
The entry standard page now shows where entries are thin, by category.
The search page shows its interface in Spanish when searching with the Spanish pilot.
Entry pages now list other entries linked to the same place or material.
Entry pages now show which other entries link to them.
Descriptor, type, category, context, source and claim pages now show their permanent identifiers.
Search results can now be narrowed to one type, such as places or collections.
Integrated nine new checks for changelog pagination, interface messages, graph health, relation exports, linked data, route coverage, the editor handbook, route security and the documentation index into the 185-check production gate.
Every public document is now indexed, and the documentation map reaches the encyclopedia, Learn and release pages.
Checked every data route is read-only, sends the right headers and publishes no personal data.
Published a practical guide for editors and volunteers, checked against the real tools.
Every machine-readable route is now in the data inventory and resource registry, and new routes cannot ship without it.
Linked data now carries permanent identifiers, places, materials and their typed links.
Each relation type is now published as its own NDJSON file.
Structural checks for the knowledge graph: orphans, dangling edges and domain and range.
Spanish page text now comes from one catalogue with English fallback, ready for more languages.
The changelog now shows the newest 50 changes, with older ones on archive pages and every old link preserved.
Integrated eight new checks for worksheets, the licence switch, entry batches, performance budgets, the map, dataset releases and classroom guides into the 176-check production gate.
Published four free lesson plans for teachers, with safety guidance and links to the Learn courses.
Every knowledge-graph release now has a sealed record of its digest, counts, commit and production deployment.
The smell map can now be filtered by place or material.
Added metallic vs mineral, resinous vs woody, marine vs salty and fermented vs tangy to Smell Vocabulary.
Measured the production build and enforced page, script and style size budgets after every build.
Editors can check several draft entries together before they enter the pipeline.
The dataset licence is now one pre-checked switch for the owner's decision; no licence has been granted.
A native speaker can confirm, correct or reject every Spanish pilot name in a spreadsheet.
Researchers can record checked citations for open questions in a spreadsheet and import them into source intake.