Workaround for optional elements in unordered groups

[lstreckeisen]

Hi all

I'm currently working on a proof of concept for a new language server for ContextMapper.
The current grammar (Xtext) for ContextMapper heavily relies on optional elements in unordered groups.
Here's an example of a grammar rule I have:

BoundedContext:
	'BoundedContext' name=ID (
		('implements' (implementedDomainParts+=[DomainPart]) ("," implementedDomainParts+=[DomainPart])*)? &
		('realizes' (realizedBoundedContexts+=[BoundedContext]) ("," realizedBoundedContexts+=[BoundedContext])*)? &
		('refines' refinedBoundedContext=[BoundedContext])?
	)
	(
		OPEN
			('domainVisionStatement' ('=') domainVisionStatement=STRING)? &
			('type' ('=') type=BoundedContextType)? &
			(('responsibilities' ('=') responsibilities+=STRING) ("," responsibilities+=STRING)*)? &
			('knowledgeLevel' ('=') knowledgeLevel=KnowledgeLevel)? &
			('implementationTechnology' ('=') implementationTechnology=STRING)? &
			('businessModel' ('=') businessModel=STRING)? &
			('evolution' ('=') evolution=Evolution)?
		CLOSE
	)?
;

Is there a way to work around the missing support for optional elements in unordered groups?
For the PoC I can go ahead and require the attributes to be in order, but for a later user release, that would be a massive breaking change I'd like to avoid.

Thanks in advance for your help!

[msujew]

Hey @lstreckeisen,

AFAIK all elements in unordered groups are optional right now. Note that we recommend to use a simple (A | B | C)* pattern with validations instead of A & B & C, as the latter usually results in unreadable parser error messages (whereas the former allows you to specify what is wrong).

[lstreckeisen]

Hi @msujew
Thanks for your fast response 😃

That is good to know, the Language Reference doesn't mention that. Might be worth adding your recommendation to the Unordered groups section there.