A quiet week on the changelog, with one useful nudge toward test hygiene
A single shipping change this week, in Tuist, gives the quarantined and flaky test pages real chronological visibility so teams can finally tell which parked tests have been sitting around too long.
It was a short week on the changelog, with only one shipped change to talk about, but it lands on a problem that has bothered me for a while: the silent accumulation of quarantined tests.
The update comes from Tuist and it is small in surface area. The quarantined tests page now exposes a sortable “Quarantined at” column, and the flaky tests page gets the parallel “Marked flaky at” column. Both timestamps are drawn from the test’s event history, so they describe the moment a test entered its current state. Re-quarantining a test starts a new period, which keeps the column honest. The timestamp always describes the same event as the “Quarantined by” attribution sitting next to it, so you are not sorting by a number that has drifted away from the person who pressed the button.
That last detail matters more than it sounds. A lot of tooling will happily let you sort by “last updated” and quietly mix in metadata edits, retries, and other noise. Here the column and the attribution are wired to the same event, which means sorting is actually doing what your eyes expect it to do.
The real point of the change is what the sort unlocks. Quarantined tests are, by definition, tests you have decided not to fix right now. They live in a kind of deferred queue: someone decided that a failing or flaky test was not blocking enough to fix this week, so they parked it and moved on. Without a chronological view, that queue just grows. Every sprint review shows the same green badge next to the build, and the parked list quietly doubles in size. Nobody notices until somebody asks, in a hurry, why the quarantine count is so high, and the answer is that half of those tests have been there for nine months.
Sorting by “Quarantined at” turns that list into something with shape. The tests that have been parked the longest float to the top, which is exactly the prompt a team needs: is this still worth quarantining, or has it rotted into something that just needs to be deleted? On the other end, sorting the other way surfaces whatever was quarantined most recently, which is the natural view when you are asking why a particular test ended up parked this week and who made the call.
The same logic applies to the flaky tests page. A flaky test that has been flaky for six months is a very different object from a flaky test that started flaking last Tuesday, and the page now lets you tell them apart at a glance.
I read this as a quiet belief statement about test infrastructure: a test’s status is not a static label, it is a position in time, and the tools we use to manage tests should make that time legible. Tuist is not the first piece of testing tooling to add timestamps to its quarantine view, but the choice to anchor the timestamp to the same event as the attribution, and to make it sortable by default, suggests the team thinks of these columns as something you will actually use every week rather than something you filter for once a quarter.
If there is a direction worth watching, it is the broader idea that test health has a half-life. Quarantine and flakiness are temporary states that are useful only if they are revisited on a schedule. Giving teams a chronological handle on those states is a precondition for any policy that says “if a test has been quarantined for more than 90 days, it needs a decision: fix, delete, or re-quarantine with a new owner.” The tooling now exists to enforce that. Whether teams actually adopt the habit is a separate question, but at least the dashboard is no longer the obstacle.
Read the full change at the Tuist changelog.
Narrated from 1 public drop.