Skip to content

Documentation describes the released artefact, and now something checks - #15

Merged
FlagshipDev merged 1 commit into
mainfrom
publish/2026-08-07-1e3df55
Aug 7, 2026
Merged

Documentation describes the released artefact, and now something checks#15
FlagshipDev merged 1 commit into
mainfrom
publish/2026-08-07-1e3df55

Conversation

@FlagshipDev

Copy link
Copy Markdown
Contributor

"Docs move with code" has been a rule here since the constitution was written, and the same defect
still happened three times: HULLWORK_ERROR_DSN and Postgres documented in three places each while
0.1.0a1 shipped without the extras that make them possible, and projects add --credential-file
documented while every document pinned an image that answers unrecognized arguments. Each sentence
was true of main and false of what a reader pulls.

A test asking the local parser whether a flag exists would have passed all three times, because all
three times the local parser was right. The gap is not between docs and code — it is between docs and
the release. So docs/published-surface.json records what the published image accepts, read by
asking that image, and twenty-one tests check the published documents against it: every command,
every flag, the four version pins, every HULLWORK_* named, and that no document is unclassified.

Caught on the real defect before it was green, and on an invented one afterwards. The rule itself,
with the four situations a contributor can be in, is in CONTRIBUTING.

Signed-off-by: Javier Miralles Rancaño 68760931+FlagshipDev@users.noreply.github.com

"Docs move with code" has been a rule here since the constitution was written, and the same defect
still happened three times: `HULLWORK_ERROR_DSN` and Postgres documented in three places each while
`0.1.0a1` shipped without the extras that make them possible, and `projects add --credential-file`
documented while every document pinned an image that answers `unrecognized arguments`. Each sentence
was true of `main` and false of what a reader pulls.

A test asking the local parser whether a flag exists would have passed all three times, because all
three times the local parser was right. The gap is not between docs and code — it is between docs and
the release. So `docs/published-surface.json` records what the published image accepts, read by
asking that image, and twenty-one tests check the published documents against it: every command,
every flag, the four version pins, every `HULLWORK_*` named, and that no document is unclassified.

Caught on the real defect before it was green, and on an invented one afterwards. The rule itself,
with the four situations a contributor can be in, is in CONTRIBUTING.

Signed-off-by: Javier Miralles Rancaño <68760931+FlagshipDev@users.noreply.github.com>
@FlagshipDev
FlagshipDev merged commit c0431d9 into main Aug 7, 2026
4 checks passed
@FlagshipDev
FlagshipDev deleted the publish/2026-08-07-1e3df55 branch August 7, 2026 08:48
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