Repository navigation
chore(coverage): stop adapter coverage from depending on packet timing - #946
Conversation
yordis
commented
Sep 28, 2026
- Master CI failed the 100% coverage gate with every test passing, because some adapter branches only ran when the server's response happened to split across packets a certain way, or when DNS answered fast enough.
- A coverage gate that flips on scheduler luck reddens unrelated merges and hides real gaps, so each of those branches now has a test that pins the packet layout it needs.
Signed-off-by: Yordis Prieto <yordis.prieto@gmail.com>
PR SummaryLow Risk Overview Finch: The streaming error test now targets a closed local listener ( Gun: Adds a test that forces Gun’s final Mint: Adds a Reviewed by Cursor Bugbot for commit f913388. Bugbot is set up for automated code reviews on this repo. Configure here. |
There was a problem hiding this comment.
Copilot review overview
🟡 Changes recommended
Gun tests do not deterministically pin packet layout, so the targeted coverage branch may remain untested.
Review effort: Lite
Findings: 1
What changed in this PR
Stabilizes adapter coverage by making packet layouts and connection failures deterministic.
Changes:
- Adds scripted Mint packet tests.
- Adds Gun multi-part response coverage.
- Uses a refused loopback connection for Finch failure testing.
| File | Reviewed changes |
|---|---|
test/tesla/adapter/mint_test.exs |
Adds deterministic packet-layout tests. |
test/tesla/adapter/gun_test.exs |
Adds multi-part body coverage, but relies on timing-sensitive sends. |
test/tesla/adapter/finch_test.exs |
Replaces DNS-dependent failure testing. |
💡 Add a code-review agent skill or configure MCP servers for context-aware, tailored reviews. Learn more in the docs.
…n TCP segmentation Signed-off-by: Yordis Prieto <yordis.prieto@gmail.com>
