Every page closed on a raw ISO stamp. It now reads data_as_of(), which
write_if_changed ignores as it ignored the old stamp. The diagram text
counts setups with setup_count(), and the data model page names the
Data and API page by its title.
generate_site.py writes the Starlight project in website/ and nothing
else: mkdocs.yml, the MkDocs hooks, theme overrides and stylesheet go,
and the wiki is written in Starlight's syntax. Both workflows set up
Node with the official actions/setup-node pinned by commit, run npm ci
from the committed lockfile, build with the strict link validator and
check the result with validate_site.py.
The search weighs a page by what it answers: Start here pages first,
a system page's cards and file names before a guide citing them, deep
guides, launchers, aliases and frozen copies after. Status badges,
hash blocks and the export change lists leave the index. The ranking
sits at the centre of a range where the 13 queries of search_gate.py
all answer, in the search box itself.
839 pages, every URL and 45,611 ids match the MkDocs build. The tests
read the Starlight content and syntax.
The pipeline locked only --output-dir while it also writes database.json,
install/, README.md and docs/. A run on a scratch output directory could
start beside the run on dist/ and rewrite them. Every run now holds the
repository's .lock first.
mkstemp creates its scratch 0600 and os.replace kept that mode, so
database.json, the manifests, the packs, the recipe snapshots and the
large-file cache were readable by their owner only. set_plain_mode
gives the scratch 0o666 less the umask, as open() would, before each
rename. A test walks every function that renames a mkstemp file.
Dropping the double-dash asides left 72 comments and docstrings with
half a dash glued to the next word ("shape -a size or a size range -is
all"). They now read as a colon, a comma or parentheses, and a prose
style test names any that comes back.
generate_pack.py held six responsibilities in 4744 lines. Five move out
in dependency order, so nothing above reaches back down: destinations,
core extras, resolution with its storage tiers, the notes shipped inside
a pack, and the pack verifier. generate_pack.py keeps the build and the
command line, and re-exports the rest.
Two things the move surfaced. The offline switch was a module global the
command line assigned, which a re-export would have copied and frozen at
False; it is set through a call now and not re-exported. And a facade
placed after the entry point binds too late: importing the module worked,
running it did not, so the manifest run died on a name the tests never
exercised because tests import.
Verified against the previous revision on frozen inputs: every generated
artefact is identical, save the catalogue that embeds hashes of files
carrying a build timestamp.
common.py had grown to 1833 lines by accumulation. Six coherent pieces
move out - untrusted parsing, digests, archives, generated artefacts,
release assets, dump catalogues - and common.py re-exports them, so the
sixty existing import sites keep working and migrating them stays
optional.
The site build is now reproducible, which is what made the move
checkable. It deleted its generated directories first, so every page was
new and write_if_changed had no earlier version to compare against: a
deploy republished six hundred pages for the clock alone. Directories
are swept instead, a page is removed only once nothing produces it, and
the body pass compares against the body of the file on disk rather than
against the decorated page. Two consecutive builds on the same inputs
now produce identical bytes; before, 1034 files differed.
Abdessamad Derraz