[Caml-list] parsing forward references in ocamlyacc and other ocamlyacc questions

[Caml-list] parsing forward references in ocamlyacc and other ocamlyacc questions

[Rafael 'Dido' Sevilla]

I'm just wondering how you can guarantee that certain imperative effects
occur, like adding a name into a symbol table hash.  I'm writing a
compiler whose semantics allow some limited forward references, e.g.:

adt_declaration: IDENTIFIER COLON ADT OBRACE adt_member_list CBRACE SEMI

where the rules under the adt_member_list nonterminal have access to the
identifier name given in the ADT declaration.  For the original Yacc or
Bison, it can be done with an intermediate semantic action like:

adt_declaration: IDENTIFIER COLON ADT { Hashtbl.add symtable $1 ... } OBRACE ...

Something like (which is how you would do it without the use of
intermediate actions):

adt_name: IDENTIFIER COLON ADT { Hashtbl.add symtable $1 adt_skel; $1 }
  ;

adt_declaration: adt_name OBRACE adt_member_list CBRACE SEMI { ... }
  ;

doesn't seem to work.  The symbol table addition still doesn't occur
until after the entire adt_declaration has been parsed.  The
documentation for ocamlyacc has very little to say about it (unless
there's more documentation in other places besides the ocaml manual).

I'll also ask how I would go about processing file inclusion in
ocamlyacc using ocamllex...

-------------------
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

Re: [Caml-list] parsing forward references in ocamlyacc and other ocamlyacc questions

[Eckart Goehler]

Hi,

On Wed, 3 Sep 2003, Rafael 'Dido' Sevilla wrote:

> I'm just wondering how you can guarantee that certain imperative effects
> occur, like adding a name into a symbol table hash.  I'm writing a
> compiler whose semantics allow some limited forward references, e.g.:
...
> until after the entire adt_declaration has been parsed.  The
> documentation for ocamlyacc has very little to say about it (unless
> there's more documentation in other places besides the ocaml manual).

As far as I understand ocamlyacc behaves as the original yacc (but not
bison).

> I'll also ask how I would go about processing file inclusion in
> ocamlyacc using ocamllex...

Tricky. Calling the parser function "main" from within the parser does not
work currently with ocamlyacc. I tried this myself and ended in patching
ocamlyacc:
- the main function generated by ocamlyacc has to be recursive (let rec
  main = ...).
- the parser table written out has to be defined before the main function.
- because ocamllex needs the token information but must be defined
  *before* the parser (otherwise you can't recursively call ocamllex) the
  token  have to be written into a different file (for foo.mly I choosed
 the output file name to be footoken.ml/footoken.mli).

But file inclusion is a bit difficult because you can't always be shure
that the processing order is kept (mentioned by the main problem you have:
imperative features while parsing.

Maybe the patch should be implemented in ocamlyacc

eckart


-------------------
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

Re: [Caml-list] parsing forward references in ocamlyacc and other ocamlyacc questions

[Michal Moskal]

On Wed, Sep 03, 2003 at 01:31:14PM +0800, Rafael 'Dido' Sevilla wrote:
> I'm just wondering how you can guarantee that certain imperative effects
> occur, like adding a name into a symbol table hash.  I'm writing a
> compiler whose semantics allow some limited forward references, e.g.:
> 
> adt_declaration: IDENTIFIER COLON ADT OBRACE adt_member_list CBRACE SEMI
> 
> where the rules under the adt_member_list nonterminal have access to the
> identifier name given in the ADT declaration.  For the original Yacc or
> Bison, it can be done with an intermediate semantic action like:
> 
> adt_declaration: IDENTIFIER COLON ADT { Hashtbl.add symtable $1 ... } OBRACE ...
> 
> Something like (which is how you would do it without the use of
> intermediate actions):
> 
> adt_name: IDENTIFIER COLON ADT { Hashtbl.add symtable $1 adt_skel; $1 }
>   ;
> 
> adt_declaration: adt_name OBRACE adt_member_list CBRACE SEMI { ... }
>   ;
> 
> doesn't seem to work.  The symbol table addition still doesn't occur
> until after the entire adt_declaration has been parsed.  The
> documentation for ocamlyacc has very little to say about it (unless
> there's more documentation in other places besides the ocaml manual).

Are you *sure*? From what I know about yacc parser, this action *had* to
be taken before adt_member_list was parsed.

-- 
: Michal Moskal :: http://www.kernel.pl/~malekith : GCS {C,UL}++++$ a? !tv
: When in doubt, use brute force. -- Ken Thompson : {E-,w}-- {b++,e}>+++ h

-------------------
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