Unhandled exceptions with --link-to-remote (type 'DynamicTypeImpl' is not a subtype of type 'InterfaceType')

[isoos]

Similar to #2101, I've found another exception that is triggered when running dartdoc with --link-to-remote:

https://pub.dev/packages/aws_sns_api

On pub.dev:
https://pub.dev/documentation/aws_sns_api/latest/log.txt

type 'DynamicTypeImpl' is not a subtype of type 'InterfaceType'
#0      TypeParameterElementType.interfaceType (package:dartdoc/src/element_type.dart:261:38)
#1      DefinedElementType.instantiatedType (package:dartdoc/src/element_type.dart:331:12)
#2      Extension._couldApplyTo (package:dartdoc/src/model.dart:1350:47)
#3      Extension.couldApplyTo (package:dartdoc/src/model.dart:1346:33)
#4      Class.potentiallyApplicableExtensions.<anonymous closure> (package:dartdoc/src/model.dart:805:27)

The fix to #2101 may have been fixed this one too, please double-check...

[jcollins-g]

Yes, this is a similar bug and already fixed.

[isoos]

Reopening this, as it was not fixed with dartdoc 0.30.0+1, example package:

• bird_flutter 0.0.3
• https://pub.dev/packages/bird_flutter
• https://pub.dev/documentation/bird_flutter/latest/log.txt

The stacktrace differs in some of the lines from the previous one:

Unhandled exception:
type 'DynamicTypeImpl' is not a subtype of type 'InterfaceType'
#0      TypeParameterElementType._interfaceType (package:dartdoc/src/element_type.dart:213:39)
#1      DefinedElementType.instantiatedType (package:dartdoc/src/element_type.dart:284:12)
#2      Extension._couldApplyTo (package:dartdoc/src/model/extension.dart:41:49)
#3      Extension.couldApplyTo (package:dartdoc/src/model/extension.dart:31:7)
#4      ExtensionTarget.potentiallyApplicableExtensions.<anonymous closure> (package:dartdoc/src/model/extension_target.dart:26:27)
#5      WhereIterator.moveNext (dart:_internal/iterable.dart:442:11)

[isoos]

(Oh, I thought I can reopen... should I create a new issue to track this?)

[jcollins-g]

This looks like some kind of different edge case. Thanks for pasting the log information into the bug, it helps a great deal!

[jcollins-g]

This is what @scheglov warned me about in a previous PR; a bug where we assume TypeParameterType.bound is an InterfaceType. Sadly it looks like it hits bugs in the real world.