Remove unneeded print statements

[talzag]

Bug/issue #, if applicable: rdar://87784021

Summary

The test LogHandleTests.testFlushesStandardOutput() fails when run outside of Xcode because stdout is written to by other tests and types. This was breaking the assumption that stdout would be empty when it was written to by the test. This PR removes several print statements from the tests that have no effect on the tests, and changes the CoverageAction type to write to a provided log handle rather than print directly to stdout.

Dependencies

None.

Testing

Steps:

1. Download the latest 5.6 toolchain snapshot from swift.org.
2. Install the toolchain and cd </path/to/swift-docc>
3. Run xcrun --toolchain org.swift.<toolchain ID> swift test (i.e. xcrun --toolchain org.swift.56202203021a swift test)
   • You can get the bundle ID of a toolchain on macOS with the command /usr/libexec/PlistBuddy -c "Print CFBundleIdentifier:" /Library/Developer/Toolchains/<toolchain>/Info.plist
4. Verify that no test failures are reported.
5. Verify that the output is formatted as expected when run with the --experimental-documentation-coverage flag (i.e. swift run docc convert --experimental-documentation-coverage --level=brief /path/to/docs.docc)

Checklist

• Added Updated tests
• Ran the ./bin/test script and it succeeded
• Updated documentation if necessary

[talzag]

@swift-ci please test

[talzag]

@swift-ci Please test Linux platform

[talzag]

@swift-ci Please test macOS platform

[daniel-grumberg]

I am confused. What has happened with the last two commits? Did you realize that change wasn't needed?

[talzag]

That’s correct. Purging stdout and stderr wasn’t necessary for the tests to pass, and fpurge is only available on BSD systems.

[daniel-grumberg]

You need to rebase this to get the latest stuff from main. While you are doing this, it would be good to remove the the purging commit and the revert of that one so the history is a bit cleaner.

[daniel-grumberg]

Apart from that looks fine.

[talzag]

@swift-ci please test

[franklinsch]

@talzag Does this resolve the issue of Swift-DocC tests failing in Swift CI during toolchain builds, and if so, why? Also, why did the test start failing in Swift CI; was it due to a change in another component in Swift?

[talzag]

@franklinsch I’m actually not sure if this is the issue that caused Swift-DocC tests to fail in the toolchain pipelines. In the radar @d-ronnqvist pointed out that none of the tests are marked as failing. However, I was seeing this test fail every time I ran the tests from the command line.

I don’t think this was caused by a change in a Swift component because I can reproduce the failure with the 5.6 snapshot toolchain and with the toolchain in my installed Xcode.

[talzag]

@swift-ci please test

[talzag]

@swift-ci please test

[icanzilb]

Hey @talzag, I'm wondering if this PR title could be improved to be more specific to the code changes it contains. The title says "ensure the stdout is empty" but I'm not sure how the code changes do that. Maybe rename to something like "Remove unneeded prints" or so? This is minutia but would be helpful when reviewing git history

[Kyle-Ye]

Encountering this bug too when I recently tested swift-docc in terminal with swift test
But strangely if I run ./bin/test, it will pass the tests.
The only difference is ./bin/test will call swift test --parallel instead of swift test
cc @talzag

[talzag]

Hey @icanzilb that’s a good point. I’ll update the title to be clearer about what this PR is changing.

[talzag]

@Kyle-Ye good catch! When tests are run in parallel, a process is spawned for each test class:

Test parallelization occurs by distributing the test classes in a target across multiple runner processes.

Each test class has it’s own stdin and stdout, so other tests can’t interfere with the LogHandleTests by writing to stdout.

[talzag]

@swift-ci please test

[talzag]

@swift-ci please test