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
Don't remove non-outline nodes that don't show up in test runs #5739
There were two different issues reported in #5734:
The test markers for tests that aren't run disappear
The test markers come back when modifying the file
I'm using that issue to track the second, but this is to track the first. It's actually by-design as the code currently is. Without the new tracking, we don't know if the test was not in the results because a) you stopped running early (or used @soloTest) or b) you had deleted the test.
However, with the new test tracking, we could perhaps remove the functionality of deleting "potentially deleted" nodes and instead just rely on the tracking to handle it (remove the node when the tracked position goes away).
It will be easier to do this when we're just using the new tracking by default, so bumping this until then,
changed the title [-]Reconsider removing "potentially deleted" tests once using new test tracking by default[/-][+]Don't remove non-outline nodes that don't show up in test runs[/+]on Dec 3, 2025
It turns out, we also need this for dynamic tests. Consider:
test_foo1() {}
We rename it:
test_foo2() {}
And then re-run the file. We need to clean test_foo1 from the results, because we do not track renames of tests (only movement of the ranges).
So this will have to stay as-is. Unfortunately this means if you run a solo-test, the markers for the other tests will still vanish, but I don't have a good solution to that that doesn't introduce other issues right now.
Activity
DanTup commentedon Dec 3, 2025
Turns out we still need this code to handle removed outline nodes, it needs to just not do result nodes.
[-]Reconsider removing "potentially deleted" tests once using new test tracking by default[/-][+]Don't remove non-outline nodes that don't show up in test runs[/+]DanTup commentedon Dec 3, 2025
It turns out, we also need this for dynamic tests. Consider:
test_foo1() {}We rename it:
test_foo2() {}And then re-run the file. We need to clean
test_foo1from the results, because we do not track renames of tests (only movement of the ranges).So this will have to stay as-is. Unfortunately this means if you run a solo-test, the markers for the other tests will still vanish, but I don't have a good solution to that that doesn't introduce other issues right now.