[release/11.0.1xx-preview7] Allow UI tests to apply resolution after test failure

[kubaflo]

Note

Are you waiting for the changes in this PR to be merged?
It would be very helpful if you could test the resulting artifacts from this PR and let us know in a comment if this change resolves your issue. Thank you!

Description of Change

The Windows UI-test resolution helper is designed to fall back to applying 1920x1080 when display enumeration or the non-mutating CDS_TEST probe is unreliable on a hosted agent. However, the generic DISP_CHANGE_FAILED probe result was treated as a hard error, so the script exited before attempting the actual resolution change.

Treat that generic probe result as a warning and continue to the real ChangeDisplaySettings call. Unsupported modes (DISP_CHANGE_BADMODE) still fail immediately, and a failed actual resolution change still fails the script. The native probe/apply result decisions are now isolated in pure helpers so those transitions can be tested without invoking Windows P/Invoke APIs.

CI Evidence

In maui-pr-uitests build 1535264, a Windows leg could not enumerate the current display and the CDS_TEST probe returned DISP_CHANGE_FAILED. The build stopped in Set screen resolution before any tests ran, even though the helper logged that it would attempt the fallback.

Testing

• Validated the updated PowerShell script with the PowerShell AST parser.
• Added 14 focused Pester 5.9 tests covering CDS_TEST decisions (FAILED continues, BADMODE rejects), successful/restart apply results, and nonzero real-apply failures.
• Ran the complete .github/scripts Pester 5.9 suite: 1,513 passed, 0 failed.

Issues Fixed

N/A

[github-actions]

🚀 Dogfood this PR with:

⚠️ WARNING: Do not do this without first carefully reviewing the code of this PR to satisfy yourself it is safe.

curl -fsSL https://raw.githubusercontent.com/dotnet/maui/main/eng/scripts/get-maui-pr.sh | bash -s -- 37044

Or

• Run remotely in PowerShell:

iex "& { $(irm https://raw.githubusercontent.com/dotnet/maui/main/eng/scripts/get-maui-pr.ps1) } 37044"

[azure-pipelines]

Azure Pipelines:
Successfully started running 1 pipeline(s).
There may be pipelines that require an authorized user to comment /azp run to run.

[Copilot]

Pull request overview

This PR adjusts the Windows screen-resolution helper script used by UI test runs so that a CDS_TEST probe failure (DISP_CHANGE_FAILED) no longer aborts the script before attempting the actual resolution change, improving resilience on hosted agents.

Changes:

• Downgrade CDS_TEST returning DISP_CHANGE_FAILED from a hard error/exit to a warning so the script proceeds to apply the requested resolution.

[kubaflo]

/azp run

[azure-pipelines]

Azure Pipelines:
Successfully started running 3 pipeline(s).

[Copilot]

Pull request overview

Copilot reviewed 1 out of 1 changed files in this pull request and generated no new comments.

[kubaflo]

/azp run maui-pr-uitests

[azure-pipelines]

Azure Pipelines:
Successfully started running 1 pipeline(s).

[PureWeen]

💡 Testing — The new DISP_CHANGE_FAILED → real-apply transition is covered only by AST parsing, so a later control-flow refactor could restore the early exit or stop propagating a failed real apply. The native static call is not mockable in the current script structure; consider a follow-up that extracts the probe-result decision into a pure helper and tests FAILED = continue and BADMODE = fail, while preserving nonzero exit for real-apply failures.
Flagged by: 3/3 reviewers after severity dispute

[kubaflo]

Agreed this is worth a follow-up. Extracting the probe-result decision (SUCCESSFUL → proceed, BADMODE → fail, FAILED → continue) into a pure helper would make the warn-and-continue transition unit-testable without the non-mockable ChangeDisplaySettings P/Invoke, and would lock in that a real-apply failure still returns nonzero.

I'm keeping it out of this PR intentionally — this change is a narrowly-scoped preview7 hot-fix (only the non-mutating CDS_TEST DISP_CHANGE_FAILED path), and restructuring the resolution flow on a Windows-only script I can't exercise on macOS is better done as its own change with the Pester coverage you describe. I'll track it as a follow-up. Leaving this thread open so it isn't lost.

[kubaflo]

Addressed in 4eb296e1ed: the CDS_TEST and real-apply result decisions are now pure helpers used by the native flow. The focused Pester suite pins FAILED = continue, BADMODE = reject, and every non-success/restart real apply = failure. All 14 focused tests and the complete 1,513-test PowerShell suite pass with Pester 5.9.0.

[PureWeen]

Adversarial review — low risk

No correctness, security, or failure-propagation issue survived consensus. The change narrowly converts only the non-mutating CDS_TEST DISP_CHANGE_FAILED result to warn-and-continue; DISP_CHANGE_BADMODE and every failed real apply still return false and produce exit code 1. The title and description accurately match the implementation.

• 💡 Testing: One non-blocking coverage suggestion was retained at the lower consensus severity. AST parsing does not exercise the new branch, but direct end-to-end stubbing would require disproportionate refactoring around static P/Invoke and unconditional process exits; a pure decision helper is the feasible follow-up.
• What looks right: The warning includes the probe result and target resolution, while the unchanged real-apply switch preserves diagnosable hard failures.
• Prior review status: The earlier warning-detail comment was addressed in c82d600; no unresolved correctness finding remains.

Methodology: 3 independent reviewers with adversarial consensus + repo domain specialist. The testing note reached 3/3 agreement only after severity dispute and is non-blocking. CI status was intentionally out of scope.

[kubaflo]

CI evidence from the exact-head UI run: maui-pr-uitests 1537883 executed 23 Set screen resolution tasks, and all 23 completed successfully. In every task, display enumeration failed and the script then successfully applied 1920x1080.

I inspected every resolution-task log. None emitted the new CDS_TEST returned DISP_CHANGE_FAILED warning, so this run supports the general fallback behavior but did not directly exercise the specific branch changed by this PR. I am leaving the existing non-blocking coverage note open and am not claiming direct proof of that branch.

[kubaflo]

/azp run

[azure-pipelines]

Azure Pipelines:
Successfully started running 3 pipeline(s).

[Copilot]

Pull request overview

Copilot reviewed 2 out of 2 changed files in this pull request and generated no new comments.

Suppressed comments (1)

.github/scripts/Set-ScreenResolution.Tests.ps1:15

• Guard against $parseErrors being $null and include just the parse error messages for consistency with other Pester AST-based tests in this repo. This avoids a potential null-reference and makes failures easier to read.

    if ($parseErrors.Count -gt 0) {
        throw "Set-ScreenResolution.ps1 has parse errors: $($parseErrors -join '; ')"
    }

[kubaflo]

@PureWeen this is ready to merge at unchanged head 4eb296e1ed. The exact synthetic merge 2533689b2f has this head as its second parent, and main build 1539091 completed successfully. The exact-head UI run also proved all 23 applicable Windows screen-resolution setup tasks, including the EnumDisplaySettings fallback successfully applying 1920x1080. Focused Pester is 14/14 and the full repository suite is 1,513/1,513. The remaining device-test failures are unrelated to this Windows UI-test infrastructure change.

[kubaflo]

Consolidated into #37081, which is now the single rebased test-only stabilization PR for Preview 7.