NewFinderTests are breaking on the build

[Ducasse]

**BBeginning to run tests of NewTools-Finder-Tests with random seed 290922244
SubscriptOutOfBounds: 1 in an OrderedCollection()
[:err |
		"Handle a drawing error"
		| errCtx errMorph |
		errCtx := thisContext.
		[
			errCtx := errCtx sender.
			"Search the sender chain to find the morph causing the problem"
			[errCtx isNotNil and:[(errCtx receiver isMorph) not]]
				whileTrue:[errCtx := errCtx sender].
			"If we're at the root of the context chain then we have a fatal drawing problem"
			errCtx ifNil:[^self handleFatalDrawingError: err description].
			errMorph := errCtx receiver.
			"If the morph causing the problem has already the #drawError flag set,
			then search for the next morph above in the caller chain."
			errMorph hasProperty: #errorOnDraw
		] whileTrue.
		errMorph setProperty: #errorOnDraw toValue: true.
		"Install the old error handler, so we can re-raise the error"
		err signal.
	] in WorldState>>displayWorldSafely: in Block: [:err |...
FullBlockClosure(BlockClosure)>>cull:
[:ex | errorHandlerBlock cull: ex] in FullBlockClosure(BlockClosure)>>onErrorDo: in Block: [:ex | errorHandlerBlock cull: ex]
FullBlockClosure(BlockClosure)>>cull:
Context>>evaluateSignal:
Context>>handleSignal:
Context>>handleSignal:
SubscriptOutOfBounds(Exception)>>signal
SubscriptOutOfBounds class>>signalFor:lowerBound:upperBound:in:
SubscriptOutOfBounds class>>signalFor:lowerBound:upperBound:
SubscriptOutOfBounds class>>signalFor:
OrderedCollection(Object)>>errorSubscriptBounds:
OrderedCollection>>at:
RubParagraph>>characterBlockForIndex:
RubShoutStylerDecorator(RubParagraphDecorator)>>characterBlockForIndex:
RubPrimarySelectionMorph(RubTextSegmentMorph)>>characterBlockForIndex:
RubPrimarySelectionMorph(RubTextSegmentMorph)>>computeStraightVertices
RubPrimarySelectionMorph(RubTextSegmentMorph)>>computeVertices
RubPrimarySelectionMorph(RubTextSegmentMorph)>>whenExtentChanged:
MessageSend>>value:
MessageSend>>cull:
MessageSend>>cull:cull:
[ action cull: anAnnouncement cull: announcer ] in AnnouncementSubscription(AbstractAnnouncementSubscription)>>deliver: in Block: [ action cull: anAnnouncement cull: announcer ]
FullBlockClosure(BlockClosure)>>on:do:
FullBlockClosure(BlockClosure)>>on:fork:
AnnouncementSubscription(AbstractAnnouncementSubscription)>>deliver:
[ subscription deliver: anAnnouncement ] in SubscriptionRegistry>>deliver:to:startingAt: in Block: [ subscription deliver: anAnnouncement ]
FullBlockClosure(BlockClosure)>>ifCurtailed:
SubscriptionRegistry>>deliver:to:startingAt:
SubscriptionRegistry>>deliver:to:
ug description**
A clear and concise description of what the bug is.

**To Reproduce**
Steps to reproduce the behavior:
1. Go to '...'
2. Click on '....'
3. Scroll down to '....'
4. See error

**Expected behavior**
A clear and concise description of what you expected to happen.

**Screenshots**
If applicable, add screenshots to help explain your problem.

**Version information:**
 - OS: [e.g. iOS]
 - Version: [e.g. 1.6.1]
 - Pharo Version [e.g 9]

**Expected development cost**
As you know nothing happens magically, so do you have an idea of the development effort required? 
In addition, are you willing to help because after all Pharo is yours too?

**Additional context**
Add any other context about the problem here.

[Rinzwind]

I guess the above is from the log of the build for pull request #16037?

The following Dockerfile can be used to reproduce the error with some additional logging:

FROM almalinux:9
# The following file is from: https://github.com/pharo-project/pharo-launcher/actions/runs/7669561590
COPY PharoLauncher-linux-37d81461-x64.zip .
RUN yum install -y unzip && yum clean all && rm -rf /var/cache/yum
RUN unzip -p PharoLauncher-linux-37d81461-x64.zip | tar -x
RUN mkdir /root/Pharo
RUN ./pharo-launcher/pharo-launcher image create fromBuild --newImageName test-image 1320
RUN /root/Pharo/vms/120-x64/pharo --headless /root/Pharo/images/test-image/test-image.image eval " \
	StFinderTest compile: 'performTest \
		Stdio stderr nextPutAll: (''Performing test (test: {1}, process: {2})'' format: { self asString. Processor activeProcess name }); lf. \
		^ super performTest'. \
	SubscriptOutOfBounds compile: 'signal \
		Stdio stderr nextPutAll: (''Signaling SubscriptOutOfBounds (process: {1})'' format: { Processor activeProcess name }); lf. \
		^ super signal'. \
	Smalltalk snapshot: true andQuit: true"
CMD /root/Pharo/vms/120-x64/pharo /root/Pharo/images/test-image/test-image.image test --junit-xml-output NewTools-Finder-Tests

Sample output with the error (after running it repeatedly as the error occurs ‘randomly’):

There was an error during copy, we should not throw an exception here:Error: Bad BitBlt arg (Fraction?); proceed to convert.
There was an error during copy, we should not throw an exception here:Error: Bad BitBlt arg (Fraction?); proceed to convert.
Running tests in 1 Packages
Performing test (test: StFinderClassTest>>#testSearchForExactCaseInsensitiveStringMissing, process: CommandLine handler process)
[…]
Performing test (test: StFinderExampleTest>>#testConcatenationSearch, process: CommandLine handler process)
Signaling SubscriptOutOfBounds (process: CommandLine handler process)
[…]
Performing test (test: StFinderSelectorTest>>#testSearchForExactCaseInsensitiveStringSucess, process: CommandLine handler process)
Signaling SubscriptOutOfBounds (process: Morphic UI Process)
Signaling SubscriptOutOfBounds (process: Morphic UI Process)
1 in an OrderedCollection()
[…]
OrderedCollection>>at:
RubParagraph>>characterBlockForIndex:
RubShoutStylerDecorator(RubParagraphDecorator)>>characterBlockForIndex:
RubPrimarySelectionMorph(RubTextSegmentMorph)>>characterBlockForIndex:
RubPrimarySelectionMorph(RubTextSegmentMorph)>>computeStraightVertices
RubPrimarySelectionMorph(RubTextSegmentMorph)>>computeVertices
RubPrimarySelectionMorph(RubTextSegmentMorph)>>whenExtentChanged:
[…]
SubscriptionRegistry>>deliver:to:startingAt:
Performing test (test: StFinderSelectorTest>>#testSearchForExactCaseSensitiveStringMissing, process: CommandLine handler process)
[…]
27 run, 27 passes, 0 failures, 0 errors.

As far as I understand, the error is due to concurrent accessing of morphs from within the command line handler and Morphic UI processes. I guess the tests should be changed so that either the window is not drawn, or the actions (#doSearch:, #setExactAndCaseInsensitive, …) are performed from within the Morphic UI process as well.

[Ducasse]

I think that your analysis is correct

[Rinzwind]

The superclass of StFinderTest should probably be changed to SpBaseTest which ensures that the tests run within the Morphic UI process, see Spec pull request #1496 and issue #12502.

A somewhat related issue is issue #9638, which was fixed in pull request #14853 by similarly deferring to the Morphic UI process.

[hernanmd]

Thanks for reporting. I couldn't reproduce it locally, but I've sent a PR pharo-spec/NewTools#669 with your suggestion.