[antlr-interest] Confusing, hopefully-final, problems

[Sam Barnett-Cormack]

Sam Barnett-Cormack wrote:
> Sam Barnett-Cormack wrote:
>> Hey all,
>>
>> With Gavin's generous help, I'm down to one error and a couple of 
>> warnings, which is nice.
>>
>> In the attached grammar, I haven't fixed the lexer issues Gavin 
>> identified yet, as I'm focused on the parser for the moment.
>>
>> The error is on line 473, as are two of the warnings. I can't say as I 
>> can figure out what it's getting at all, but I'm probably just being dumb.
>>
>> The rule is
>>
>> extensionAndException : '...' exceptionSpec? ;
>>
>> and exceptionSpec is:
>>
>> exceptionSpec : ('!' exceptionIdentification)? ;
>>
>> exceptionIdentification : signedNumber | definedValue | type ':' value ;
>>
>> The error and warnings are:
>>
>> warning(200): workspace/asn.1-parse/grammars/ASN_1.g:346:82: Decision 
>> can match input such as "'('" using multiple alternatives: 1, 2
>> As a result, alternative(s) 2 were disabled for that input
>> warning(200): workspace/asn.1-parse/grammars/ASN_1.g:473:31: Decision 
>> can match input such as "'}'" using multiple alternatives: 1, 2
>> As a result, alternative(s) 2 were disabled for that input
>> warning(200): workspace/asn.1-parse/grammars/ASN_1.g:473:31: Decision 
>> can match input such as "','" using multiple alternatives: 1, 2
>> As a result, alternative(s) 2 were disabled for that input
>> error(201): workspace/asn.1-parse/grammars/ASN_1.g:473:31: The following 
>> alternatives can never be matched: 2
>>
>> My best guess is that the problem comes from "type" in the 
>> exceptionIdentification rule, but then I'd expect the error to be there, 
>> and not in extensionAndException.
> 
> No, Sam, it's because a rule is referenced as optional, and that rule 
> can match nothing as well, so there's no point it being optional...
> 
> So I'm now down to a warning about
> 
> type : ((typeWithConstraint)=>typeWithConstraint | builtinType | 
> referencedType) constraint* ;
> 
> And a load of parser warnings and errors that it wasn't bothering to 
> generate before... *sigh*

Or even *lexer* warnings and errors

Sam