[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: Grammar troubles for the Eiffel Language (cont'd, cont'd)

The problem is that your grammar is ambiguous.

You wrote:

	class_declaration =
		...
		features?
		...
		;

...

	features =
		{empty} |
		{nonempty} feature feature_clause? features
		;

So, if you omit "features" in an input text, the parser does not know if (1) you have omitted "features", or (2) if you have included an "empty features".

So, you must chose.  Either you write
  ... | features | ...
or you get rid of the empty alternative of features.

:-)

Etienne

Dave Makalsky wrote:
Hi Etienne, others,

The Eiffel grammar is nearly complete... however I am getting a shift / reduce
conflict, and I can't understand why.

I have included my grammar file.

Please advise.

Cheers,

Dave Makalsky

Error Message:

shift/reduce conflict in state [stack: PClassHeader *] on TInvariant in {
        [ PFeatures = * ] followed by TInvariant (reduce),
        [ PInvariant = * TInvariant PAssertion ] (shift)
}
java.lang.RuntimeException:

shift/reduce conflict in state [stack: PClassHeader *] on TInvariant in {
        [ PFeatures = * ] followed by TInvariant (reduce),
        [ PInvariant = * TInvariant PAssertion ] (shift)
}
        at org.sablecc.sablecc.Grammar.computeLALR(Unknown Source)
        at org.sablecc.sablecc.GenParser.caseStart(Unknown Source)
        at org.sablecc.sablecc.node.Start.apply(Unknown Source)
        at org.sablecc.sablecc.SableCC.processGrammar(Unknown Source)
        at org.sablecc.sablecc.SableCC.processGrammar(Unknown Source)
        at org.sablecc.sablecc.SableCC.main(Unknown Source)


Quoting Etienne Gagnon <etienne.gagnon@uqam.ca>:


Hi Dave,

Dave Makalsky wrote:

Hi:

I am moving along nicely with my Eiffel grammar, but I hit a small snag.

I have the following definition:

feature_adaptation =
 rename?  new_exports? undefine? redefine? select? end

How do I allow the keyword 'end' in the feature_adaptation iff at least one

of

the optional constructs are present?

I assume feature_adaptation is *used* with a "?" when it apprears somewhere else in the grammar (e.g. feature_adaptation?).

There are 2 usual ways. The first is to delay the detection of the error in a post-parsing walk of the AST.

The second is to write:

feature_adaptation =
  rename new_exports? undefine? redefine? select? end |
  new_exports undefine? redefine? select? end |
  undefine redefine? select? end |
  redefine select? end |
  select end;

Using SableCC 3, this can be converted back into a nice AST, e.g.

  rename new_exports? undefine? redefine? select? end
    {-> New feature_adaptation (rename, new_exports, undefine,
                                redefine, select) |
  new_exports undefine? redefine? select? end
    {-> New feature_adaptation (null, new_exports, undefine,
                                redefine, select) |
  ...

I assume the AST does not need to keep the end token around.

Etienne

--
Etienne M. Gagnon, Ph.D.             http://www.info.uqam.ca/~egagnon/
SableVM:                                       http://www.sablevm.org/
SableCC:                                       http://www.sablecc.org/



-------------------------------------------------------
This sf.net email is sponsored by:ThinkGeek
Welcome to geek heaven.
http://thinkgeek.com/sf
_______________________________________________
Sablecc-user mailing list
Sablecc-user@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/sablecc-user






----------------------------------------------------------------
This message was sent using IMP, the Internet Messaging Program.



--
Etienne M. Gagnon, Ph.D.             http://www.info.uqam.ca/~egagnon/
SableVM:                                       http://www.sablevm.org/
SableCC:                                       http://www.sablecc.org/