Repository navigation
Conversation
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Taking the app samples out of the platform solution filters meant msbuild.exe no longer built the MAUI sample, or any other Android or Windows app. The Samples job still builds it with the .NET CLI. The MSBuild job finishes well before the slowest build leg, so it builds the sample again, after the solution filter. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The win-arm64 runner has 4 cores. With the unit tests in the background next to the integration tests, Pester took 876.6 s on the combined draft, against 684 s with the unit tests in the foreground, and aot.Tests went from 31 s to 2.6 minutes. That made the leg the slowest check, at 30 to 32 minutes. The macOS MAUI job becomes a matrix with a win-arm64 entry. It builds and packs only the packages the integration tests use, then runs all of them, so the leg keeps the build, the unit tests and the coverage upload. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Without a solution filter, the environment action restores the workloads for every project in Sentry.slnx. That took 184 s in the win-arm64 integration test job, which only builds 5 projects. The action now takes the job's project files, and skips the Android test app's workloads for them, as it does for a solution filter without the test app. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
…samples job The Samples job on win-x64 still installed workloads as MSI packages under Program Files, 163 s, and restored the MAUI sample from the NuGet cache in the user profile, 180 s. It now uses the same drive step as the other Windows jobs. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The environment action now requires the solution filters or project files the job builds, and installs only their workloads. Jobs that build everything pass Sentry.slnx. Before, every job without a filter restored the workloads for all of Sentry.slnx, and CodeQL, format-code and verify-api ran a second, unfiltered workload restore after the action. The Playwright job restores the workloads of its test project only. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The win-arm64 leg got 14 minutes faster when its integration tests moved to their own job, because its 4 cores couldn't run them next to the unit tests. The win-x64 and linux-x64 runners have 4 cores too, and their legs were the slowest after that, so they get entries in the integration-test matrix as well. The legs keep the build, the unit tests and the unsupported AOT test. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
…API approval job It was the last Windows job without the drive step. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Every Windows job installs the same .NET SDKs next to the workspace, and the arm64 ones the Android SDK, which their image lacks. That took 1 to 3.5 minutes per job. The environment action now restores both from one cache per architecture, keyed on global.json and the action itself, and skips the installs on a hit. The workload restore still runs, so each job adds whatever its projects need that the cached copy lacks. The build legs save the cache on a miss, because they install the most workloads. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Codecov Report✅ All modified and coverable lines are covered by tests. Additional details and impacted files@@ Coverage Diff @@
## main #5805 +/- ##
=======================================
Coverage 76.13% 76.13%
=======================================
Files 528 528
Lines 19579 19579
Branches 3829 3829
=======================================
Hits 14906 14906
Misses 3791 3791
Partials 882 882 ☔ View full report in Codecov by Harness. 🚀 New features to boost your workflow:
|
ric-oliv
deleted the
ci/measure-integration-splits-and-toolchain-cache
branch
October 25, 2026 12:16
This was referenced Oct 25, 2026
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Throwaway draft to measure three more changes on top of #5804's state. I'll close it after three runs and fold each change into the PR it belongs to.
What I'm checking:
#skip-changelog