Re: [Caml-list] eliminating shift/reduce conflicts

["Eric C. Cooper" <ecc@cmu.edu>]

On Fri, Sep 12, 2003 at 04:26:32PM +0800, Rafael 'Dido' Sevilla wrote:
> I have an ocamlyacc grammar that contains productions that look like:
> 
> foo_list: FOO { [$1] }
>   | foo_list COMMA FOO { $3 :: $1 }
>   ;
> 
> which creates a list of FOO objects.  I however have some rules that
> need to be prefixed by either a single FOO or a foo_list, like so:
> 
> bar: foo_list COLON xyzzy { ... }
> 
> and
> 
> baz: FOO COLON yzzyx { ... }
> 
> This of course produces a shift/reduce conflict, and shifting fails to
> parse the 'bar' correctly.  Perhaps I need to read a compiler
> construction textbook more thoroughly to figure out this answer, but any
> hints out there on getting rid of this shift/reduce.

Here are two approaches.  If the grammar is simple enough, you can
split the bar rule (and other similar ones) into two productions:

    bar: FOO COLON xyzzy | foo_list COLON xyzzy

and require foo_list to have at least two members:

    foo_list: FOO COMMA FOO | foo_list COMMA FOO

Of course, now you have duplicated semantic actions for the two cases,
and it can make the grammar pretty ugly.

Another approach is to accept a more liberal language in the parser,
and do more checking in the semantic action:

    bar: foo_list COLON xyzzy { ... }
    baz: foo_list COLON yzzyx { check_single_foo $1; ... }

Hope this helps.

-- 
Eric C. Cooper          e c c @ c m u . e d u

-------------------
To unsubscribe, mail caml-list-request@inria.fr Archives: http://caml.inria.fr
Bug reports: http://caml.inria.fr/bin/caml-bugs FAQ: http://caml.inria.fr/FAQ/
Beginner's list: http://groups.yahoo.com/group/ocaml_beginners