Re: [s-ramp] Fwd: Explanation of s-ramp query grammar problem

[Vince Brunssen <brunssen@us.ibm.com>]

Kurt,

Can you please open an issue on this and we will address it at the meeting next week?

Thank you!

Regards,
Vince Brunssen
Senior Software Engineer
Emerging Technologies
512-286-5487 (External)
363-5487 (T/L)

Kurt T Stam ---07/18/2012 09:39:34 AM---Hi Guys, Eric works with me on S-RAMP and while implementing the Query API he

From:	Kurt T Stam <kstam@redhat.com>
To:	"s-ramp@lists.oasis-open.org" <s-ramp@lists.oasis-open.org>,
Cc:	Eric Wittmann <eric.wittmann@redhat.com>
Date:	07/18/2012 09:39 AM
Subject:	[s-ramp] Fwd: Explanation of s-ramp query grammar problem
Sent by:	<s-ramp@lists.oasis-open.org>

---

Hi Guys,

Eric works with me on S-RAMP and while implementing the Query API he noticed that our query grammar does not supports literals to be arguments in function calls. Would it make sense to update the grammer slightly?

Thx,

--Kurt


-------- Original Message --------

Subject:	Explanation of s-ramp query grammar problem
Date:	Tue, 17 Jul 2012 10:23:58 -0400
From:	Eric Wittmann <eric.wittmann@redhat.com>
To:	Kurt Stam <kstam@redhat.com>

Problem:

As defined, the S-RAMP query grammar does not allow literals to be
arguments to function calls.


Relevant Grammar:

FunctionCall ::= FunctionName '(' ( Argument ( ',' Argument )* )? ')'
Argument ::= Expr
Expr ::= EqualityExpr
EqualityExpr ::= ForwardPropertyStep '=' PrimaryExpr
PrimaryExpr ::= Literal  |  Number  |  '$' PropertyQName

I condensed the grammar a bit (took out boolean _expression_ definitions
and whatnot) to illustrate the point.  You can see that the only way to
get to a PrimaryExpr is on the right-hand side of an EqualityExpr.


Example:

/s-ramp/xsd/XsdDocument[classifiedByAnyOf(., 'taste')]

The example above breaks down as an artifact set with a predicate where
the predicate is a function call with two arguments.  The second
argument is expected to be PrimaryExpr:

ArtifactSet [ FunctionCall ( RelationshipPath, PrimaryExpr ) ]

But the grammar states that an Argument must be an Expr, which does
*not* allow it to be a simple PrimaryExpr.


Proposal:

Update the grammar, changing Argument to be defined as:

Argument ::= PrimaryExpr
          | Expr

This is the simplest grammar change that will support the example, but
it is not the only one.