From marcoxa at cs.nyu.edu  Mon Apr 16 19:44:54 2007
From: marcoxa at cs.nyu.edu (Marco Antoniotti)
Date: Mon, 16 Apr 2007 21:44:54 +0200
Subject: [cl-unification-devel] Re: [CL-UNIFICATION] MATCH,
	MATCHING macro's bug?
In-Reply-To: <4dadc9270704161000x7ba030ev687f67c988a0a710@mail.gmail.com>
References: <4dadc9270704161000x7ba030ev687f67c988a0a710@mail.gmail.com>
Message-ID: <71B9566C-F61D-4AF8-9840-9D90F3FB79DC@cs.nyu.edu>

Hi Masayuki

it looks like the patch is good.

Let me do a few tests and I will check it in.

Apart from that, thanks for using CL-UNIFICATION.

Cheers

Marco





On Apr 16, 2007, at 7:00 PM, Masayuki Onjo wrote:

> Hello, Marco.
>
> I'm Masayuki Onjo, CL-UNIFICATION library user. The CL-UNIFICATION is
> very cool stuff.
> It's works well on AllegroCL's REPL, but COMPILE-FILE cause  
> following error.
>
> -- fib.lisp ---
> (defun fib (n)
>  (unify:match-case (n)
>    (0 1)
>    (#T(number n)
>       (+ (fib (- n 1)) (fib (- n 2))))))
>
>  CL-USER> (compile-file "fib.lisp")
>
>  Object #<EMPTY UNIFY ENVIRONMENT: 1 frame @ #x22b212d2> cannot be
>  written to a compiled file unless an applicable method is defined on
>  make-load-form.
>     [Condition of type PROGRAM-ERROR]
>
> It seems to be a bug of the MATCH macro expansion.
>
>  [BUG]
>   (unify ',template ,object ,substitution))
>     substitution := (make-empty-environment)
>     => (unify temlate object #<EMPTY UNIFY ENVIRONMENT: 1 frame @  
> #...>)
>
>  [CORRECT]
>   (unify ',template ,object ,substitution))
>     substitution := '(make-empty-environment)
>     => (unify temlate object (make-empty-environment))
>
> I make following patch, it works for me. I'm very happy to hacking
> with pattern-match-based macro.  Thanks for your great work!!
>
> cvs diff: Diffing .
> Index: match-block.lisp
> ===================================================================
> RCS file: /project/cl-unification/cvsroot/cl-unification/match- 
> block.lisp,v
> retrieving revision 1.6
> diff -r1.6 match-block.lisp
> 13c13
> <                            (substitution (make-empty-environment))
> ---
>>                            (substitution '(make-empty-environment))
> 63c63
> <                           (default-substitution (make-empty- 
> environment)))
> ---
>>                           (default-substitution '(make-empty- 
>> environment)))
>
> -- 
> Masayuki Onjo <onjo at lispuser.net>

--
Marco Antoniotti


-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://mailman.common-lisp.net/pipermail/cl-unification-devel/attachments/20070416/2868929f/attachment.html>

From marcoxa at cs.nyu.edu  Wed Apr 18 09:08:48 2007
From: marcoxa at cs.nyu.edu (Marco Antoniotti)
Date: Wed, 18 Apr 2007 11:08:48 +0200
Subject: [cl-unification-devel] Re: [CL-UNIFICATION] MATCH,
	MATCHING macro's bug?
In-Reply-To: <4dadc9270704171535i38e859ffi5e0765f8327c8041@mail.gmail.com>
References: <4dadc9270704161000x7ba030ev687f67c988a0a710@mail.gmail.com>
	<71B9566C-F61D-4AF8-9840-9D90F3FB79DC@cs.nyu.edu>
	<4dadc9270704171535i38e859ffi5e0765f8327c8041@mail.gmail.com>
Message-ID: <02128D56-241F-41A4-B296-76AD70CAEFE5@cs.nyu.edu>

Hi Masayuki,

thanks.  Yep.  You are right.

Please stand by for a few days before these changes make it in the  
codebase.

Cheers

Marco




On Apr 18, 2007, at 12:35 AM, Masayuki Onjo wrote:

> Hi, Marco
>
>> it looks like the patch is good.
>>
>> Let me do a few tests and I will check it in.
>
> Thanks!
>
> I'm enjoying UNIFICATION power. The pattern matching is fun. :-)
> There are some more bugs in CONTROL-FLOW macros.
>
> CL-USER> (macroexpand-1 '(unify:match-case (x) (0 1) (1 2) (2 3)
> (otherwise -1)))
> (LET ((#:OBJECT-VAR-3066 X))
>  (HANDLER-CASE (CL.EXT.DACF.UNIFICATION:MATCH (0 #:OBJECT-VAR-3066) 1)
>                (CL.EXT.DACF.UNIFICATION::UNIFICATION-FAILURE NIL
>                 (HANDLER-CASE
>                  (CL.EXT.DACF.UNIFICATION:MATCH (1 #:OBJECT- 
> VAR-3066) 2)
>                  (CL.EXT.DACF.UNIFICATION::UNIFICATION-FAILURE NIL
>                   (HANDLER-CASE
>                    (CL.EXT.DACF.UNIFICATION:MATCH (2 #:OBJECT- 
> VAR-3066) 3)
>                    (CL.EXT.DACF.UNIFICATION::UNIFICATION-FAILURE NIL
>                     (OTHERWISE -1))))))))
>
> I think (OTHERWISE -1) must be (PROGN -1) and
>
> CL-USER> (macroexpand-1 '(unify:matching () ((0 x) 1) ((1 y) 2)  
> (otherwise -1)))
> (BLOCK CL.EXT.DACF.UNIFICATION:MATCHING
>  (COND
>   ((SETF #:UNIFICATION-ENV-3070
>            (IGNORE-ERRORS
>             (CL.EXT.DACF.UNIFICATION:UNIFY '0 X
>
> (CL.EXT.DACF.UNIFICATION:MAKE-EMPTY-ENVIRONMENT))))
>    (LET* ()
>      1))
>   ((SETF #:UNIFICATION-ENV-3071
>            (IGNORE-ERRORS
>             (CL.EXT.DACF.UNIFICATION:UNIFY '1 Y
>
> (CL.EXT.DACF.UNIFICATION:MAKE-EMPTY-ENVIRONMENT))))
>    (LET* ()
>      2))
>   (NIL
>    (ERROR 'CL.EXT.DACF.UNIFICATION::UNIFICATION-NON- 
> EXHAUSTIVE :FORMAT-CONTROL
>           "Non exhaustive matching."))
>   (OTHERWISE -1)))
>
> (OTHERWISE -1) should be (T -1), too. I made following patch.
>
> Index: match-block.lisp
> ===================================================================
> RCS file: /project/cl-unification/cvsroot/cl-unification/match- 
> block.lisp,v
> retrieving revision 1.6
> diff -r1.6 match-block.lisp
> 13c13
> <                            (substitution (make-empty-environment))
> ---
>>                            (substitution '(make-empty-environment))
> 63c63
> <                           (default-substitution (make-empty- 
> environment)))
> ---
>>                           (default-substitution '(make-empty- 
>> environment)))
> 139c139
> <              ,@(when default-clause (list default-clause)))))
> ---
>>              ,@(when default-clause (list `(T ,(cdr default- 
>> clause)))))))
> 191c191
> <                  otherwise-clause
> ---
>>                  `(progn ,(cdr otherwise-clause))
>
> -- 
> Masayuki Onjo <onjo at lispuser.net>

--
Marco Antoniotti


-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://mailman.common-lisp.net/pipermail/cl-unification-devel/attachments/20070418/ff09d62e/attachment.html>