Rascal type-checker crashed unexpectedly with: No module scope found for analysis::diff::TextEditsat: |lib://rascal-core/lang/rascalcore/check/Import.rsc|(9426,2,<234,57>,<234,59>)

[DavyLandman]

Context:

module A

import analysis::diff::TextEdits;

Exception thrown:

|lib://rascal-core/lang/rascalcore/check/Import.rsc|:234,57: "No module scope found for analysis::diff::TextEdits"

Stacktrace:

	at getModuleScope(|lib://rascal-core/lang/rascalcore/check/Import.rsc|(9047,384,<225,0>,<235,1>))
	at prepareForCompilation(|lib://rascal-core/lang/rascalcore/check/Import.rsc|(10886,38,<263,17>,<263,55>))
	at rascalTModelForLocs(|lib://rascal-core/lang/rascalcore/check/Checker.rsc|(16482,77,<385,38>,<385,115>))
	at check(|lib://rascal-core/lang/rascalcore/check/Checker.rsc|(22699,64,<545,9>,<545,73>))
	at $shell$(|main://$shell$|)

[DavyLandman]

Okay, this message appears when the module doesn't exist. it should have been analysis::diff::edits::TextEdits

[DavyLandman]

@PaulKlint maybe this should not be an exception, but an error reported via the summary? since we know for which import this holds?

[PaulKlint]

This is a long overdue improvement that looks simpler than it actually is. Reason: the natural place to check for a non-existing module is the collect rule for import/extend. However, the modular structure of the Rascal program is analyzed long before those rules are reached. Therefore non-existing modules could create havoc during this analysis. The module existence check is now implemented in that early fase, hopefully making the checker much more robust.

[PaulKlint]

After testing, I will push this fix.