|
| From: | Ramaswamy |
| Subject: | GLR parsing problem |
| Date: | Thu, 13 Feb 2003 17:51:56 +0530 |
|
Hi,
I have a problem with
merging in bisons GLR parsing mechanism. The following is the grammar (which btw
I did not design this grammar, someone else is to blame and I dun have enough
time to redesign the grammar) where I am facing some problem (I have shown only
the relavent portion).
Assignment: ObjectAssignment | ValueAssignment ObjectAssignment:
LC_IDENTIFIER TYPE_REFERENCE ASGN ObjectDefn
%merge<Merge>
ValueAssignment :
LC_IDENTIFIER TYPE_REFERENCE ASGN Value %merge<Merge>
Value: SequenceOfOrSetOfValue
%merge<cacheAmbiguity>
Value: CharacterStringValue
%merge<cacheAmbiguity>
Unfortunately ObjectDefn,
SequenceOfOrSetOfValue and CharacterStringValue all can parse one particular structure
of input. So I decide to %merge them but the order in which the Merge fn is called is not very clear to me. Since there r 3
ways to parse the input 3 stacks exist and at the end since all of them end at
the same state state 1 and 2 are merged with stack 0. Thus I expect 2 calls
to Merge fn. But I got this sequence for the
input " { 1, 2 } " -
<<SequenceOrSetOfValue>>
<<CharacterStringValue>> Merge( SequenceOfOrSetOfValue , CharacterStringValue0 ) -> SC0 <<ValAssgn:SC0>> <<CharacterStringValue1>> <<ValAssgn:CharacterStringValue1>> Merge( SC0 , CharacterStringValue1 ) -> SC1 <<ObjAssgn:9:ObjectDefn>> Merge( SC1 , ObjectDefn ) From my understanding I expected the
second and 3rd part of the sequence to be somethin like -
<<SequenceOrSetOfValue>>
<<CharacterStringValue>> Merge( SequenceOfOrSetOfValue , CharacterStringValue0 ) -> SC0 <<ValAssgn:SC0>> <<ObjAssgn:ObjectDefn>> Merge( SC0 , ObjectDefn ) -> SC1 Cud any one could explain why
this is happening or cud give reference to some detailed use of %merge. Thankin
in advance.
Regds
Ram |
| [Prev in Thread] | Current Thread | [Next in Thread] |