You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
{{ message }}
Repository navigation
Private key test strings block PRs that touch the reviewer's own tests #35
A PR that touches review/tests/unit/test_collect.py or test_render.py gets non-waivable incomplete-coverage hard blocks. The cause is the test strings that hold a -----BEGIN ... PRIVATE KEY----- header. They are built with Python +, so the code after the header is not key material, and rule R-KEYBLOCK marks the file unavailable. This happened on #34.
Splitting the header strings in our tests would stop the blocks. But a split header is also a way any author can avoid the header check today. So first check what the detection covers.
Questions
How does the prefilter detect a committed private key: the header only, or the key body too?
Does the threat model count an author who commits a key on purpose and splits the header? That author has easier ways, such as base64 encoding.
Is there a gap to fix in detection, separate from our own test files?
Then decide: split the headers in our tests, move the key fixtures, or keep the admin bypass for PRs that touch these files.
Not in scope
Changing what the gate blocks without a spec update.
Activity