Consolidate grammar for Synonyms on Attributes and Enums

[regnosys]

Consolidate grammar for Synonyms on Attributes and Enums

[dhuebner]

@jim-h-wang

1. I'm for using double-quotes. Comment fron the Grammar the synonymValue is of type STRING to accommodate peculiar values which is true i.e. value "ACT/360". This will also help keep ValidId exceptional cases reduced, see #101

2. Also here I would say use quotes. There is no benefit for . notation, one can not use content assist after . and one can't jump to the path segments.

3. see 2.

Afaik we do not really have control over "referenced" values in synonym part, correct me if I'm wrong. If so, we can't really react on syntactically breaking cases for example an arithmetical operator, reserved keyword or even space. Overall, the only practical way I see, is to quote the values.

[jim-h-wang]

@dhuebner

1. Ok, let's use quotes everywhere for consistency.

2, 3. Sounds like you are in favour of migrating . to ->. For this change, you will need to find usages of the path part of the synonym across all repositories. I know at least in rosetta-ingest, we depend on the . to perform a String::split

[dhuebner]

@jim-h-wang

depend on the . to perform a String::split

Hopefully some tests will fail if I miss a piece

[dhuebner]

@jim-h-wang
I would also suggest to consolidate the rest of class synonym rule:

RosettaSynonymBody:
	( 'value'  (values+=RosettaSynonymValue ',')* values+=RosettaSynonymValue (mappingLogic=RosettaMapping)? 
		| 'hint' (hints+=ValidHint ',')* hints+=ValidHint
		| mappingLogic=RosettaMappingSetTo )
	('meta' (metaValues += ValidID ',')* metaValues += ValidID)?
	('mapper' mapper=ValidID)?
	
;

ValidHint & ValidID are just values which are not referenced or referencing things hance may become a STRING too. Also ValidHint allows * at the end but it's never used in cdm, do we still need this?

I found any usage of SynonymEnumeration:

RosettaSynonymEnumeration:
	value = '.' enumValue=STRING
;

One test case for that, but the syntax looks really strange:

[synonym FpML value Bar."value"]

[jim-h-wang]

ValidHint & ValidID are just values which are not referenced or referencing things hance may become a STRING too. Also ValidHint allows * at the end but it's never used in cdm, do we still need this?

I still prefer to not have the quotes if we can avoid them. Changing ValidId to STRING would mean that all the attribute names would then need quotes. The * isn't currently used but we do still plan to make use of it.

I found any usage of SynonymEnumeration:

We can remove this grammar, it's unused in the model and not supported in ingestion.