Skip to content

Correct the release caution before tagging v0.1.1 - #4

Merged
Boernsman merged 1 commit into
mainfrom
release-notes-truth
Oct 10, 2026
Merged

Boernsman merged 1 commit into
mainfrom
release-notes-truth

Conversation

@Boernsman

Copy link
Copy Markdown
Contributor

The generated release notes open with a [!CAUTION] block stating that the hardware adapters kiln_hal_esp32s3 do not exist yet. They do, and the hardware configuration has been built by CI on every push since the first build of it turned up two -Wformat-truncation errors.

Tagging v0.1.1 against the current file would publish that sentence as a safety statement about a kiln controller, so this goes in first.

The corrected block is both accurate and less comfortable than the old one:

  • the published image is still the simulated configuration and drives no outputs;
  • the hardware configuration builds, so the adapters are no longer absent;
  • none of it has run against a board — no thermocouple front end, no SSR, no contactor, no current transformer, no supervisor link, no display;
  • the charge-pump release of SWA-05, the central safety property of the design, can only be verified on hardware.

The hardware image stays out of the release, and the notes now say why: an image that looks flashable and has never driven a contactor is the most dangerous thing this repository could offer. The --prerelease comment is corrected on the same point — "built" is not the bar for dropping it, "run" is.

Workflow-only change; yaml.safe_load parses it. After this merges, v0.1.1 gets tagged on main.

The generated notes open with a CAUTION block, which is the one part of a
release anybody is obliged to read, and it says the hardware adapters
"do not exist yet". They do: kiln_hal_esp32s3 is in the tree, the hardware
configuration builds, and CI has built it on every push since the day that
the first build of it turned up two -Wformat-truncation errors.

Tagging v0.1.1 against this file would have published that sentence as a
safety statement about a kiln controller, so it is corrected before the tag
rather than after it.

What the block says now, which is both true and less comfortable: the
published image is still the simulated one and drives no outputs, the
hardware configuration builds, and **none of it has run against a board** --
no thermocouple front end, no SSR, no contactor, no current transformer, no
supervisor link, no display. The charge-pump release of SWA-05, the central
safety property of the design, cannot be verified anywhere but on hardware.

The hardware image stays out of the release deliberately, and the notes now
say why: an image that looks flashable and has never driven a contactor is
the most dangerous artefact this repository could offer. "Built" is not the
bar for dropping --prerelease either; "run" is, and the comment there says so.
@Boernsman
Boernsman merged commit 43a4c6c into main Oct 10, 2026
11 checks passed
@Boernsman
Boernsman deleted the release-notes-truth branch October 10, 2026 10:16
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