Skip to content

fix(ilttu): run ilttu-design v0.1.1, built from main - #5

Merged
ajbeck merged 1 commit into
mainfrom
ajb/pin-ilttu-design-v0.1.1
Nov 15, 2026
Merged

ajbeck merged 1 commit into
mainfrom
ajb/pin-ilttu-design-v0.1.1

Conversation

@ajbeck

@ajbeck ajbeck commented Nov 15, 2026

Copy link
Copy Markdown
Contributor

For asteroid-computing/ilttu#733: re-cut ilttu-design from main after #3's squash merge, and repin the wrapper, so every pinned binary is built from reviewed main.

What changed

The tag ilttu-design-v0.1.1 is on 401b686, #3's squash commit on main, cut as v0.1.0 was, an annotated tag that the release workflow built and published.

Verified

  • The four published binaries match the published SHA256SUMS, and each pinned sum is that file's.
  • Each binary is stamped vcs.revision=401b6865987fb8fde5e764fd2a30e390f39e8036, and version prints ilttu-design-v0.1.1.
  • A build on this Mac from a clean checkout of 401b686, with the workflow's own steps and Go 1.27.4, makes all four binaries byte for byte.
  • The wrapper downloads v0.1.1 into an empty data folder, runs it, and runs it again from the cache; its cached binary has the pinned sum.
  • claude plugin validate ., claude plugin validate plugins/ilttu, and go vet and go test in ilttu-design/.

A flaw in the release build, for #4

Three of the four binaries say vcs.modified=true; darwin-arm64 says false.
The workflow builds into ilttu-design/dist/, inside the checkout, so after the first build Go sees untracked files and marks the next three modified.
The bytes are still 401b686's alone, as the reproduction above shows; only the flag is wrong, and v0.1.0 had the same flaw, which is why only its darwin-arm64 build reproduced before.
#4's release workflow builds the same way, so I'm fixing it there, to build outside the checkout.

Not verified

  • The linux and darwin-amd64 binaries running; they were only reproduced and checksummed.

Pins the wrapper to ilttu-design-v0.1.1, cut from 401b686 on main, and
each binary's SHA-256 from its published SHA256SUMS. Bumps the plugin to
0.3.1.

Refs asteroid-computing/ilttu#733
@ajbeck

ajbeck commented Nov 15, 2026

Copy link
Copy Markdown
Contributor Author

Ready for review: #5, for asteroid-computing/ilttu#733, head 4283014a88d0a53149d1255387efeec174a63f97.

What it is

  • Tag ilttu-design-v0.1.1 is an annotated tag on 401b686, cut as v0.1.0 was, after AJ confirmed the push in this session; the release workflow built and published it.
  • The PR pins the wrapper to v0.1.1 and its four SHA-256s, and bumps the plugin to 0.3.1.

Verified

  • The four published binaries match the published SHA256SUMS, and each pin is that file's sum.
  • Each binary is stamped vcs.revision=401b6865987fb8fde5e764fd2a30e390f39e8036, and version prints ilttu-design-v0.1.1.
  • A build on this Mac from a clean checkout of 401b686, with the workflow's steps and Go 1.27.4, makes all four binaries byte for byte.
  • The wrapper downloads v0.1.1 into an empty data folder, runs it, and runs it from the cache.
  • claude plugin validate on the marketplace and the plugin, go vet and go test.

Not verified

  • The linux and darwin-amd64 binaries running; they were reproduced and checksummed only.

For AJ to know

@ajbeck ajbeck left a comment

Copy link
Copy Markdown
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Review of 4283014: approve

The PR repins the wrapper to ilttu-design-v0.1.1 and bumps the plugin to 0.3.1, for asteroid-computing/ilttu#733, as AJ approved in #733's Decided.

Verified:

  • The tag ilttu-design-v0.1.1 points to 401b686, PR #3's squash-merge commit on main.
  • Each of the four pins equals the digest GitHub records for that release asset: darwin-arm64 e51bc8c8…, darwin-amd64 008ec1dd…, linux-arm64 314715d7…, linux-amd64 4bd856bb….
  • The diff changes only the tag, the four pins and the plugin's version.

For AJ to know: three of the four binaries are stamped vcs.modified=true, because the release workflow builds inside the checkout. Their bytes are still 401b686's, as the session's byte-for-byte reproduction shows. v0.1.0 had the same flaw. PR #4's commit 0251a4b fixes the workflow, so no re-cut is needed.

Not verified:

  • I didn't download or run the binaries; the pins were checked against GitHub's asset digests, not files I hashed.
  • The wrapper's cold download.
  • PR #4, which is reviewed on its own.

@ajbeck
ajbeck merged commit ffd0703 into main Nov 15, 2026
@ajbeck
ajbeck deleted the ajb/pin-ilttu-design-v0.1.1 branch November 15, 2026 16:29
ajbeck added a commit that referenced this pull request Nov 15, 2026
Start release-please's history at #5's merge, so the plugin's first
release PR doesn't repeat what 0.3.1 shipped. Fail the build step when
git status fails, download SHA256SUMS in a step of its own so the pin
script's step has no token in its env, and let a rerun of the pin job
succeed once the pin is on main.
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