fivsevn-devlog

System map

Content and build source for devlog.fivsevn.com. Jekyll renders the note modules; a Python build overlays Fieldlog’s static pages into the same GitHub Pages artifact.

Content and publishing

Path Responsibility
index.md Homepage artwork and links
posts/ Year/month publication packages, each with index.md and its own media or attachments
natsci/, netcom/ Topic-organized Markdown notes, translations and generated module directories
natsci/fieldlog/ Dated Fieldlog packages, image upload entry, publishing state and guide
intake/ News intake sources, generated intake page and its validation
blogops/ Manually maintained blog operations records
now/, system-map.md Current state and repository map
stills/, motion/, savour/ Entry guides to the repositories that own those boards
scripts/, schemas/, tests/ Build tools, shared metadata copies and checks

Edit notes in their module and push to main. First-level _drafts/ notes are handled by the existing metadata completion workflow and do not independently trigger a deployment. The module-index updater preserves its declared manual sections. Do not use it to rewrite the operations log.

Before each Fieldlog upload, sync the latest main in Working Copy or GitHub Desktop. Save the shortcut’s Images into this repository’s full natsci/fieldlog/trigger_imgs/ path, then commit and push to main. Each input becomes one editable package with optional commit-description text. The importer saves a WebP, publishes and verifies it, then clears only matching inbox originals. Saved Fieldlog titles, captions and publication controls can be edited in index.md.

Daily publishing adds normal commits and does not rewrite Git history. History cleanup is off by default and runs only through manual workflow_dispatch with compact_image_history=true. Keep original images and unpushed work backed up outside the repository. After that maintenance, realign or re-clone old local checkouts; do not merge the old history.

Build and Actions

Build and deploy devlog checks out current main and the authoritative spec, normalizes and validates Fieldlog, imports pending uploads, refreshes module directories, runs Jekyll, overlays Fieldlog, and verifies source and generated links. It saves generated content and state together, deploys through GitHub Pages, verifies public entry points and image bytes, then clears only verified inbox originals.

The intake workflow compiles intake/sources.md with intake/config.yml, writes index.html and llm-context.md directly, and triggers a Pages rebuild after a successful run. The draft, intake and metadata workflows have separate responsibilities and remain active. Pages uses _config.yml, the existing layouts, and the custom domain in CNAME; Fieldlog must not disable Jekyll for the note modules.

python -m pip install -r scripts/requirements-fieldlog.txt -r intake/requirements.txt
python -m unittest discover -s tests -v
python -m unittest discover -s scripts -p 'test_fieldlog*.py' -v
python scripts/site_post_metadata.py check --site fieldlog --root natsci/fieldlog --schema-dir schemas
python scripts/media_naming.py check --site fieldlog --root natsci/fieldlog
python scripts/build_fieldlog.py
python scripts/build_fieldlog.py --check

The full Jekyll artifact is built and checked by Actions. scripts/check_devlog_site.py checks the combined website; scripts/verify_fieldlog_live.py compares Fieldlog’s deployment receipt with the public pages and media.

The spec repository defines common frontmatter, the site-post profile and publication media names. Local helper/schema copies keep content checks offline. Both workflows read the pinned revision from .github/spec-ref; upgrade this pin and the local helper/schema copies together. Keep IDs and publication instants stable when editing saved packages; hidden and draft packages are omitted from public output.