Skip to content

Put the hashes where someone will look for them - #38

Merged
NiKrause merged 1 commit into
mainfrom
ci/release-summary
Sep 19, 2026
Merged

NiKrause merged 1 commit into
mainfrom
ci/release-summary

Conversation

@NiKrause

Copy link
Copy Markdown
Collaborator

You asked why the run page carries no summary. It does not, and it should — the hashes exist to be compared against a local build of the same tag, and they were going into an annotation plus twenty thousand lines of wine output.

Now they go into $GITHUB_STEP_SUMMARY, which is the first thing on the run page:

Windows win32 — dc4.1.7

file bytes sha256
electrum-doi-dc4.1.7-portable.exe 39260904 2a96c10f48056db373bf4574f4bfb2deae82c27b675d61e6b8af608cbd793121
electrum-doi-dc4.1.7-setup.exe 29169608 92a15df2fd55566886b2deee6d8f34126235defe76d745fcd0c3e407c47c2cb5
electrum-doi-dc4.1.7.exe 39259320 b7690c4ddd189714f636e138b8e952ccbe801dd978b43e6701b972f6e7aef31b

Built from a clean checkout on a runner. A local build of the same tag should produce the same three hashes; if it does not, something is in one tree that is not in the other.

Those are the real numbers — the block was cut out of the workflow and run against the released 4.1.7 binaries, so stat -c%s and sha256sum are confirmed to agree with what the release carries rather than assumed to.

The heading falls back to the dispatch input when git describe cannot answer. git is on the runner, but an empty code span is a silly way to discover it is not.

The ::notice annotation goes away: it concatenated all three hashes onto one line, which was hard to read and harder to compare.

🤖 Generated with Claude Code

They exist to be compared against a local build of the same tag, and they
were going into an annotation and 20,000 lines of wine output. Now they go
into the run's summary page as a table with file, size and sha256, under a
heading naming the ref that was built.

Checked by cutting the block out of the workflow and running it against the
4.1.7 binaries: the table renders, stat -c%s and sha256sum agree with what
the release carries. The heading falls back to the dispatch input when
`git describe` cannot answer, because an empty code span is a silly way to
discover git is missing.

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
@NiKrause
NiKrause merged commit b78bd70 into main Sep 19, 2026
4 checks passed
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