Skip to content

Bump dotnet-stryker to 5.0.0 and give the full mutation run five hours - #120

Merged
Pixnop merged 2 commits into
devfrom
chore/stryker-5.0.0
Oct 2, 2026
Merged

Pixnop merged 2 commits into
devfrom
chore/stryker-5.0.0

Conversation

@Pixnop

@Pixnop Pixnop commented Oct 2, 2026

Copy link
Copy Markdown
Contributor

Takes Dependabot's bump of dotnet-stryker to 5.0.0 (#115) with the one thing it needs to be judged, a full run, and the one thing that run showed: the lane's time limit is too short on either version.

Same scores on both versions

Two full runs on the same commit and the same day, one per version:

4.16.0 5.0.0
Pulse: score 94.48 % 94.48 %
Pulse: killed, timed out, survived 685, 0, 30 681, 4, 30
Pulse: duration 3 h 33 min 3 h 18 min
Pulse.Otlp: score 81.00 % 81.00 %
Pulse.Otlp: killed, survived 179, 29 179, 29
Pulse.Otlp: duration 14 min 7 min

Both versions create the same 1,454 and 514 mutants and test the same 715 and 208. Four mutants that 4.16.0 killed are caught by a timeout on 5.0.0, which counts the same; the survivors are as many. The thresholds (83 and 70) are far below both.

Of the two breaking changes of 5.0.0, one is already met (it targets the .NET 10 runtime, which the lane installs) and the other does not apply (the baseline provider, which the lane does not use).

The full run gets five hours

The first full run on 5.0.0 was cancelled at the lane's limit of 150 minutes and reported no score. It was not the new version: the 4.16.0 run of the same day needed 3 h 33 min. At the 0.2.0 release the run took 2 h 15 min, fifteen minutes under the limit. The limit of the full run goes from 150 to 300 minutes; the pull request run keeps its 20.

The weekly run reads this file from main, where the limit is still 150: that needs the same change there.

Closes #115.

Pixnop added 2 commits October 2, 2026 19:52
The full run of Pulse, 715 mutants, took 2 h 15 min at the 0.2.0 release against a limit of 150
minutes, and 3 h 18 min and 3 h 33 min in two runs on 2026-10-02: at 150 the run is cancelled and
reports no score. 300 minutes leaves room for a slow runner. The pull request run keeps its 20
minutes.
Two full runs on the same commit and the same day, one per version, give the same scores: Pulse
94.48 % on both (4.16.0: 685 killed, 30 survived; 5.0.0: 681 killed, 4 timed out, 30 survived) and
Pulse.Otlp 81.00 % on both (179 killed, 29 survived). Four mutants that 4.16.0 killed are caught by
a timeout on 5.0.0, which counts the same. Pulse took 3 h 18 min on 5.0.0 against 3 h 33 min on
4.16.0, and Pulse.Otlp 7 min against 14.

5.0.0 targets the .NET 10 runtime, which this lane already installs. Its other breaking change,
the baseline provider following the output path, does not apply: the lane uses no baseline.
@Pixnop
Pixnop merged commit 461b07c into dev Oct 2, 2026
8 checks passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant