There are quite a few tests that expect PASS, TIMEOUT, and NOTRUN outcomes (or some subset) at the test and subtest level, but notFAIL or CRASH outcomes. The key observation here is we haven't actually observed these to not PASS; I'll call these "not-not PASSing tests". My hypothesis is that many of these will simply be PASSing when we run those tests in tier 2's broader timeout window. We should triage not-not PASSing tests with a new heuristic for update-backlog.
There are quite a few tests that expect
PASS
,TIMEOUT
, andNOTRUN
outcomes (or some subset) at the test and subtest level, but notFAIL
orCRASH
outcomes. The key observation here is we haven't actually observed these to notPASS
; I'll call these "not-notPASS
ing tests". My hypothesis is that many of these will simply bePASS
ing when we run those tests in tier 2's broader timeout window. We should triage not-notPASS
ing tests with a new heuristic forupdate-backlog
.