Skip to content

release: spec-kit c6bbf0c to main - #47

Merged
jm-rivera merged 13 commits into
mainfrom
release/2026-11-01
Nov 1, 2026
Merged

jm-rivera merged 13 commits into
mainfrom
release/2026-11-01

Conversation

@jm-rivera

Copy link
Copy Markdown
Contributor

Releases spec-kit at c6bbf0c to main, cut on release/2026-11-01: five pull requests.

dfo-apis' release pins this commit, so this merges first.

Testing: CI passed on each pull request into dev and on c6bbf0c.

…ount-base

feat(units): declare the count base of a per-N rate
dfo-dev-apis refuses anonymous callers, so `record` against dev answered
403 on every request. `record --cloud-run-token` mints the gcloud
account's Google ID token, sends it in `X-Serverless-Authorization` on
each request to the base URL's origin, and mints a new one every 45
minutes. Signed storage URLs get no token, and no record holds it.
… apis ccb661792

served/openapi.json comes from dfo-apis ccb661792, and every recorded
scenario from its dev deploy. The document gains the analysis, derived,
export, frame and share routes (14 paths, 71 schemas) and
EnrichmentOptionModel.in_long_layout, true when a long-layout download
writes the enrichment with no columns token.

Search now binds the ODA disbursement questions to net flows, so
ref-one-member follows fetch-spec-party-totals-kept to its Net spec,
version and file. The long builds moved to 1968 and 1969.
A card or dataset version description can now state the publisher's
own indicator codes behind its series (World Bank IDS's
DT.DIS.BLAT.CB.CD and the like) as about.source_codes, and each drawn
value its variables declare can carry its own. A code is
{provenance, code, exact?} in Search's order and with its meaning of
exact. A list holds 1 to 50 codes and is left out past that, as
drawn.values is. A computed version and a source card take none, and
a card drawn from a computed version takes no exact code. The
components catalogue manifest carries the schema's new digest.
uniqueItems compares whole entries, so a source_codes list could name
the same (provenance, code) twice when only exact differed, and could
list codes in any order. check_description now refuses a list under
about or on a drawn value that is not strictly ascending by
(provenance, code), with the reason source_codes_not_ascending, which
covers both. Two vectors pin a duplicate on a drawn value and a
reversed about list.
…d-in-long-layout

feat(client-contract): recapture the served document and re-record at apis ccb661792
…ption-source-codes

feat(descriptions): carry each series' publisher codes
@jm-rivera
jm-rivera merged commit 53ef70a into main Nov 1, 2026
38 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