`ReifiedTripleSubject`, `TripleTermSubject`, `TripleTermDataSubject` and `ExprTripleTermSubject` rules allow literals but not triple terms

[Tpt]

The ReifiedTripleSubject grammar rule allow literals but forbids triple terms (TripleTerm rule). The same thing happens in TripleTermSubject and ExprTripleTermSubject

I tend to think that we should:

1. either make reified triple patterns fully symmetric and allow triple terms
2. either make reified triple patterns non symmetric and not allow literals

The current median state seems weird to me

[afs]

I'm not against making a change, but there is a reason to the current state.

Literals are allowed in triple patterns in SPARQL (from 1.0) - that's a fixed point.
It allows for inverse property paths.
While not relevant to inside triple terms, it is how triple patterns are, and the doc makes inside and outside the same.

Our choice at 1.2 is whether to allow triple terms or not in the subject position.
The doc currently follows Turtle given the long debates about the issue.

I prefer no change or option 1.

[Tpt]

Literals are allowed in triple patterns in SPARQL (from 1.0) - that's a fixed point.

Yes for triple patterns! What I am talking about are the grammar rules affecting the content of triple terms, not the rules affecting "regular" triple patterns from SPARQL 1.1.
In more concrete terms, I am suggesting to change

[117]   	ReifiedTripleSubject 	  ::=   	Var | iri | RDFLiteral | NumericLiteral | BooleanLiteral | BlankNode | ReifiedTriple
[120]   	TripleTermSubject 	  ::=   	Var | iri | RDFLiteral | NumericLiteral | BooleanLiteral | BlankNode

into (with triple terms, option 1)

[117]   	ReifiedTripleSubject 	  ::=   	Var | iri | RDFLiteral | NumericLiteral | BooleanLiteral | BlankNode | ReifiedTriple | TripleTerm
[120]   	TripleTermSubject 	  ::=   	Var | iri | RDFLiteral | NumericLiteral | BooleanLiteral | BlankNode | TripleTerm

or (no literal anymore, option 2)

[117]   	ReifiedTripleSubject 	  ::=   	Var | iri | BlankNode | ReifiedTriple
[120]   	TripleTermSubject 	  ::=   	Var | iri | BlankNode

Both grammar rules are new into RDF 1.2. I find it weird that we have literals allowed inside of triple terms in subject position but not triple terms.

[domel]

+1 for option 1

[Tpt]

Update: ExprTripleTermSubject has also this discrepency

[afs]

And TripleTermDataSubject.