Fix Reversed Logic for QuantaUpkeepInstant.now() #223
Conversation
|
Whooooops! Good catch, thanks! |
governor/src/clock/quanta.rs
Outdated
| @@ -100,8 +100,7 @@ impl Clock for QuantaUpkeepClock { | |||
|
|
|||
| fn now(&self) -> Self::Instant { | |||
| QuantaInstant(Nanos::from( | |||
| self.reference | |||
| .saturating_duration_since(self.clock.recent()), | |||
| self.clock.recent().saturating_duration_since(self.reference), | |||
There was a problem hiding this comment.
Can you re-run cargo fmt and commit the change, please?
There was a problem hiding this comment.
No problems, reformatted and combined the new advance test as only one Upkeep thread can be running at a time and this seemed the simplest way to avoid issues.
Thought there might have been a solution to running multiple tests against Upkeep in the Quanta repo, but they only have one too. Didn't catch it the first time around as I use nextest and it had these tests isolated.
|
OK, this is truly weird: running the tests in "coverage" mode, it sounds like the clock isn't advancing! https://github.com/boinkor-net/governor/actions/runs/7919482513/job/21626813726?pr=223#step:5:246 |
|
I just decreased the frequencies to see if its an issue with the background task not running fast enough. I did have the interval set to 10µs, which is relatively fast. |
Codecov ReportAttention:
Additional details and impacted files@@ Coverage Diff @@
## master #223 +/- ##
==========================================
- Coverage 98.25% 98.18% -0.08%
==========================================
Files 31 31
Lines 2182 2148 -34
==========================================
- Hits 2144 2109 -35
- Misses 38 39 +1 ☔ View full report in Codecov by Sentry. |
|
OK, that looks better, now there's just one tiny amount of formatting to fix up ((: |
|
I really need to add auto-formatting on save to my IDE! 😅 |
|
I don't understand that codecov failure… it's pointing at the actual error message in the test! |
1ef6f50
|
Oh yeah, that's a problem in rcov; it counts line-broken expressions as separate lines, and since this is a macro in which the formatter arguments aren't evaluated unless the tests fail, they count as "uncovered". It's silly & unless we redo the macro as a single line, it's unavoidable. The change merges regardless! (: |
Logic for comparing
self.referenceandself.clock.recent()was reversed resulting innow()always returningNanos(0).