When using Sunshine to parse the following Entity instance (it ends with a single empty line, but YellowGrass doesn’t show that):

module example
  entity URL {
}a     

I get this error:

java.lang.NullPointerException
09:58:56.912 [main] FATAL org.metaborg.sunshine.CompilerCrashHandler - Exiting because of previous error
at org.metaborg.sunshine.model.messages.CodeRegionHelper.weaveDamageLines(CodeRegionHelper.java:30)
at org.metaborg.sunshine.model.messages.CodeRegion.getDamagedRegion(CodeRegion.java:119)
at org.metaborg.sunshine.model.messages.Message.toString(Message.java:91)
at java.lang.String.valueOf(String.java:2854)
at java.lang.StringBuilder.append(StringBuilder.java:128)
at org.metaborg.sunshine.model.messages.MessageEmitter.emitMessages(MessageEmitter.java:20)
at org.metaborg.sunshine.drivers.SunshineMainDriver.run(SunshineMainDriver.java:201)
at org.metaborg.sunshine.drivers.Main.main(Main.java:50)

I also had similar cases where the error is at the end of the line, but not necessarily at the end of the file.

I tried to debug it by adding print statements to reportErrorNearOffset(..):

private void reportErrorNearOffset(ITokenizer tokenizer, int offset, String message) {
	IToken errorToken = tokenizer.getErrorTokenOrAdjunct(offset);
	System.out.println(offset + errorToken);
	reportErrorAtTokens(errorToken, errorToken, message);
}

For the Entity instance, this prints 30 but I would expect 29a or 30a (not sure how to count the tokens).
This suggests that it is an issue with JSGLR. However, the Eclipse editor does report the errors correctly.
Perhaps there a hack is provided in Spoofax Eclipse to workaround the JSGLR issue? Or perhaps this is not an issue with JSGLR but an issue with Sunshine?

Submitted by Oskar van Rest on 13 August 2014 at 19:28