Skip to content

docs: record that 0.2.1 was yanked, and what a yank does not do - #113

Open
tschm wants to merge 2 commits into
mainfrom
docs/provenance-yank-status
Open

docs: record that 0.2.1 was yanked, and what a yank does not do#113
tschm wants to merge 2 commits into
mainfrom
docs/provenance-yank-status

Conversation

@tschm

@tschm tschm commented Aug 23, 2026

Copy link
Copy Markdown
Member

PROVENANCE.md said:

0.2.1 remains on PyPI and still carries it; anyone auditing that specific
artefact should know what they are looking at.

That was written before the release was yanked, and now reads as though nothing
was done about it. It is the one document in the repo whose entire purpose is to
be an accurate contemporaneous record, so a stale sentence costs more there than
it would anywhere else — and it is the page you would point an outside auditor at.

Verified against PyPI rather than from memory:

0.2.1: yanked=True (2/2 files)  reason={'GPL-derived test'}
0.2.2: yanked=False
0.3.0: yanked=False
0.3.1: yanked=False
0.4.0: yanked=False

The replacement records the yank and its reason, and keeps the distinction that
matters rather than claiming more than was done: a yank stops every resolver from
selecting the version, but it is not deletion, and cvx-quadprog==0.2.1 still
installs. An auditor needs both halves of that.

Documentation only — no source, no tests, no published figures touched.

🤖 Generated with Claude Code

PROVENANCE.md said "0.2.1 remains on PyPI and still carries it", which was
written before the release was yanked and now reads as though nothing was done
about it. That is the one document in the repo whose entire purpose is to be an
accurate contemporaneous record, so a stale sentence there costs more than it
would anywhere else.

Verified against PyPI rather than from memory: both files of 0.2.1 are yanked,
with `GPL-derived test` as the stated reason, and no later release is.

The replacement keeps the distinction that actually matters instead of claiming
more than was done. A yank stops every resolver from selecting the version; it is
not deletion, and `cvx-quadprog==0.2.1` still installs. Anyone auditing that
specific artefact needs to know both halves of that.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Copilot AI lite review requested due to automatic review settings August 23, 2026 11:23

Copilot AI left a comment

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

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

Pull request overview

Updates PROVENANCE.md to accurately reflect the current PyPI status of release 0.2.1 (now yanked), and clarifies what yanking does and does not imply for auditors reviewing historical artifacts.

Changes:

  • Replaces stale wording about 0.2.1 “remaining on PyPI” with an explicit record that it was yanked and why.
  • Clarifies that yanking affects default resolver selection but does not delete the artifact or prevent exact pins from installing it.

💡 Add a code-review agent skill or configure MCP servers for context-aware, tailored reviews. Learn more in the docs.

Comment thread PROVENANCE.md
Comment on lines +70 to +72
0.2.1 has since been **yanked**, with `GPL-derived test` as the stated reason, so
no resolver will select it. Yanking is not deletion, and the distinction is the
point of recording it here: the artefact is still on PyPI and an exact pin
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.

2 participants