PROBE (do not merge): does UTS_SIDE reach the test process on simulators?

[maratal]

Throwaway, to be closed as soon as it has answered its question.

Sets Test/UTSDevice.xctestplan's configuration value to bogus and restricts the run to SideSeamTests.

Side.swift aborts on an unrecognised UTS_SIDE, so the two outcomes are unambiguous:

• device legs crash with Fatal error: UTS_SIDE must be 'core' or 'device', got 'bogus' — the plan delivers the variable on that platform.
• device legs pass — the variable never arrived, the leg silently ran as core, and the device coverage on that platform is illusory.

The core legs are the control and should pass either way.

This is the only way to tell, because a device leg that receives nothing runs as core and passes exactly like one that works.

[coderabbitai]

Important

Draft PR not reviewed

Draft PRs are not automatically reviewed by default.

• Trigger a manual review

To automatically review draft PRs, update your CodeRabbit configuration:

reviews:
  auto_review:
    drafts: true

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands.}

[maratal]

Answered: the plan does deliver UTS_SIDE on simulators.

Leg	Result
iOS … device, UTSDevice	failed with Fatal error: UTS_SIDE must be 'core' or 'device', got 'bogus'
tvOS … device, UTSDevice	same
macOS … device, UTSDevice	cancelled by fail-fast; already proven locally
all three core, Ably	passed, as the control

So the device legs on #2268 are running genuine device-mode coverage on all three platforms.

This probe was the only way to establish that: a device leg that receives nothing runs as core and passes exactly like one that works, so a green leg proves nothing. That is the same trap that made TEST_RUNNER_ appear to work.

Closing; nothing here is meant to merge.