Repository navigation
A fresh install takes Audubon's illustrations from the open dataset, not the mirror (W-816) - #378
Merged
wr merged 2 commits intoOct 20, 2026
Conversation
…not the mirror (W-816) The plates-v1 release in this repo was deleted in September, so every install has fetched the Havell plates one by one from nathanbuchar's mirror again. The open dataset already publishes every Havell sheet with its sha256 (wr/historical-bird-plates, havell-v2), byte for byte the files the mirror serves (434 of 435 match a cached copy; 165 is the Havell plate on purpose), so fetch_plates now asks that release first and keeps a sheet only when it matches the manifest, then the mirror, then audubon.org. The plate list comes from the dataset's havell/plates.csv, pinned at tag v1.2.0, and is cached beside data.json as havell-plates.csv: data.json stays the mirror's own, since export_dataset corrects its shifted names itself. An install that already has data.json keeps using it, with no request. Both URLs sit in havell.yaml's folio header, as the Gould folios' release does, and the Havell and Gould paths share one checked download (release_file). The packing path into this repo (plate_release.py, pack_plates.py, make plates-pack) is gone: there is no release here to fill. Refs: W-816
…-816) Plate 165's file is named by HAVELL_FIXES, so its slug never matches the mirror's title; with the dataset's corrected titles it was the only plate the name check still flagged on a fresh install. Refs: W-816
wr
deleted the
wells/w-816-publish-the-plate-release-again-plates-v2-from-the-purged
branch
October 20, 2026 04:05
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
What changed
The
plates-v1release in this repo was deleted in September, so every install has fetched the Havell plates one by one from nathanbuchar's mirror again. The open dataset already publishes every Havell sheet with its sha256 (havell-v2in wr/historical-bird-plates), byte for byte the files the mirror serves: 434 of 435 match a cached copy, and 165 differs on purpose (the Havell plate, W-943).fetch_plates.pyasks that release first for each Havell plate and keeps a sheet only when it matches the release'smanifest.json, then the mirror, then audubon.org. A release it cannot reach costs one request, not one per plate.havell/plates.csv, pinned at tagv1.2.0, cached besidedata.jsonashavell-plates.csv.data.jsonstays the mirror's own, becauseexport_datasetcorrects that file's shifted names itself. An install that already hasdata.jsonkeeps using it, with no request.havell.yaml's folio header (release,plates_csv), as the Gould folios'releasealready does, and the Havell and Gould paths share one checked download (release_file).plate_release.py,pack_plates.py,make plates-pack): there is no release here to fill.THIRD_PARTY.mdnow names the dataset's release.How it was verified
make test: 1405 passed, 2 skipped. New tests: release first and checked, a wrong checksum falls through to the mirror, an unreachable release costs one request, the plate list from the dataset and its fallback to the mirror, the folio header's URLs.FEATHERFRAME_PLATES_MIRROR_URLand audubon.org both pointed at a dead port. All 400 curated plates (431 species) came down in 251 s, from github.com and raw.githubusercontent.com only, with no request to the dead hosts. Every file matches the manifest, and 165 is stored as the Havell plate.What is left
Nothing here. The wiki's Sibling projects page still credits the mirror as where the illustrations come from, which stays true as provenance.
Refs W-816