jscheme-user

[Jscheme-user] treating #null as #f in conditionals?

[Ken A. <kan...@bb...>]

Here's a proposal.

I think (and i know Rusty really does) think JScheme handling of #null is awk weird.  What if in conditionals we treated #null as #f.  This would make code more scheme like, here are 2 examples.

#null is currently treated at #t in conditionals but this change may not break much code.


(define (default  value default)
  ;; Before 
  (if (isNull value) default
      value))

(define (default value default)
  ;; After.
  (or value default))

(define (memoize key table computation)
  ;; Before.
  (let ((it (.get key table)))
    (if (not (isNull it)) it
	(let ((it (computation key)))
	  (.put table key it)
	  it))))

(define (memoize key table computation)
  ;; After.
  (or (.get key table)
      (let ((it (computation key)))
	  (.put table key it)
	  it)))

Re: [Jscheme-user] treating #null as #f in conditionals?

[Geoffrey K. <ge...@kn...>]

As a practical matter, it certainly seems convenient to be able to 
treat #null the same as #f in conditionals--I agree it is 
counterintuitive to have #null be not-false.

But I'm concerned about breaking R5RS 6.3.1., "only #f counts as false 
in conditional expressions,"

http://www.swiss.ai.mit.edu/~jaffer/r5rs_8.html#SEC58

note:  (not '())  ==>  #f

and 3.2, "no object satisfies more than one of the following 
predicates: boolean? pair? symbol? number? char? string? vector? port? 
procedure?"

http://www.swiss.ai.mit.edu/~jaffer/r5rs_5.html#SEC20

So we'd have to caveat the language standard, and surprise newcomers to 
JScheme from other Schemes.  On the principle of least surprise, we'd 
probably be 50-50 between surprising newcomes to JScheme from other 
languages vs. other Schemes.

In Java you can't do:  Object o = NULL; if (o) { true-stuff} else { 
false-stuff }
you have to do:  if (o == NULL) ...
It's a pain in the butt, but it's been drilled into us.

Reluctantly, I suggest leaving #null and the semantics of `if' alone, 
and offering a "new" `if', e.g., `jif', that would treat #null as #f.

By the way, I tried to figure out what #null is in JScheme, and it 
isn't anything:

 > (symbol? #null)
#f
 > (boolean? #null)
#f
 > (pair? #null)
#f
 > (vector? #null)
#f
 > (procedure? #null)
#f
 > (string? #null)
#f
 > (port? #null)
(port? '#null)
   ====================================
SchemeException: ERROR: undefined variable "port?"
 > (number? #null)
#f
 > (char? #null)
#f

Then I remembered that the empty list '() isn't anything either, but it 
also isn't #null:

 > (eq? '() #null)
#f
 > (eqv? '() #null)
#f
 > (equal? '() #null)
#f

Geoffrey
--
Geoffrey S. Knauth | http://knauth.org/gsk

On Nov 3, 2004, at 16:32, Ken Anderson wrote:

> Here's a proposal.
>
> I think (and i know Rusty really does) think JScheme handling of #null 
> is awk weird.  What if in conditionals we treated #null as #f.  This 
> would make code more scheme like, here are 2 examples.
>
> #null is currently treated at #t in conditionals but this change may 
> not break much code.
>
>
> (define (default  value default)
>   ;; Before
>   (if (isNull value) default
>       value))
>
> (define (default value default)
>   ;; After.
>   (or value default))
>
> (define (memoize key table computation)
>   ;; Before.
>   (let ((it (.get key table)))
>     (if (not (isNull it)) it
> 	(let ((it (computation key)))
> 	  (.put table key it)
> 	  it))))
>
> (define (memoize key table computation)
>   ;; After.
>   (or (.get key table)
>       (let ((it (computation key)))
> 	  (.put table key it)
> 	  it)))

Re: [Jscheme-user] treating #null as #f in conditionals?

[Michael T. <mt...@bb...>]

I'm not sure I agree:  what is wrong with defining #f and #null as 
semantically equivalent?  Even if #null printed as #f, it might be 
better than the current situation.  Bottom line is that #null is a 
projection of a non-scheme thing into scheme space - it is up to jscheme 
to manage that projection in whatever way makes the most sense.  What if 
we got rid of #null altogether?... or make #null a variable bound to #f.

cheers,
    -mik

P.S.  what bugs me is having to do (equal? foo #null):
 > (or (null? #null) (not #null))
#f

Geoffrey Knauth wrote:

> As a practical matter, it certainly seems convenient to be able to 
> treat #null the same as #f in conditionals--I agree it is 
> counterintuitive to have #null be not-false.
>
> But I'm concerned about breaking R5RS 6.3.1., "only #f counts as false 
> in conditional expressions,"
>
> http://www.swiss.ai.mit.edu/~jaffer/r5rs_8.html#SEC58
>
> note:  (not '())  ==>  #f
>
> and 3.2, "no object satisfies more than one of the following 
> predicates: boolean? pair? symbol? number? char? string? vector? port? 
> procedure?"
>
> http://www.swiss.ai.mit.edu/~jaffer/r5rs_5.html#SEC20
>
> So we'd have to caveat the language standard, and surprise newcomers 
> to JScheme from other Schemes.  On the principle of least surprise, 
> we'd probably be 50-50 between surprising newcomes to JScheme from 
> other languages vs. other Schemes.
>
> In Java you can't do:  Object o = NULL; if (o) { true-stuff} else { 
> false-stuff }
> you have to do:  if (o == NULL) ...
> It's a pain in the butt, but it's been drilled into us.
>
> Reluctantly, I suggest leaving #null and the semantics of `if' alone, 
> and offering a "new" `if', e.g., `jif', that would treat #null as #f.
>
> By the way, I tried to figure out what #null is in JScheme, and it 
> isn't anything:
>
> > (symbol? #null)
> #f
> > (boolean? #null)
> #f
> > (pair? #null)
> #f
> > (vector? #null)
> #f
> > (procedure? #null)
> #f
> > (string? #null)
> #f
> > (port? #null)
> (port? '#null)
>   ====================================
> SchemeException: ERROR: undefined variable "port?"
> > (number? #null)
> #f
> > (char? #null)
> #f
>
> Then I remembered that the empty list '() isn't anything either, but 
> it also isn't #null:
>
> > (eq? '() #null)
> #f
> > (eqv? '() #null)
> #f
> > (equal? '() #null)
> #f

and, of course,

 > (null? #null)
#f

As an alternative, what about defining #f and #null as the same object?

Re: [Jscheme-user] treating #null as #f in conditionals?

[Timothy J. H. <ti...@cs...>]

On Nov 4, 2004, at 2:50 PM, Michael Thome wrote:

> I'm not sure I agree:  what is wrong with defining #f and #null as 
> semantically equivalent?
I would side with Rusty and Geoffrey.

One problem with identifying #f and #null is that they have different 
semantics in Java
and hence would create problems when calling Java methods/constructors 
etc.
e.g. to create a Boolean array of size 4 and store two Boolean values 
in it, we could use

   >   (define f (list->array Boolean.class  '( #f #t #null #null)))

But if #f and #null were identified, we would not be able to represent 
#null



>   Even if #null printed as #f, it might be better than the current 
> situation.
Again, it would create problems when working with the Java interface. 
You would like to know
when a Boolean function is returning #null or #f.

>  Bottom line is that #null is a projection of a non-scheme thing into 
> scheme space - it is up to jscheme to manage that projection in 
> whatever way makes the most sense.
True, but we want the Scheme space to embed into Java space as well.


>   What if we got rid of #null altogether?... or make #null a variable 
> bound to #f.
>
> cheers,
>    -mik
>
> P.S.  what bugs me is having to do (equal? foo #null):
> > (or (null? #null) (not #null))
> #f
I agree with Rusty that macros would provide a mechanism for getting 
this kind of behavior.

>
> Geoffrey Knauth wrote:
>
>> As a practical matter, it certainly seems convenient to be able to 
>> treat #null the same as #f in conditionals--I agree it is 
>> counterintuitive to have #null be not-false.
>>
>> But I'm concerned about breaking R5RS 6.3.1., "only #f counts as 
>> false in conditional expressions,"
>>
>> http://www.swiss.ai.mit.edu/~jaffer/r5rs_8.html#SEC58
>>
>> note:  (not '())  ==>  #f
>>
>> and 3.2, "no object satisfies more than one of the following 
>> predicates: boolean? pair? symbol? number? char? string? vector? 
>> port? procedure?"
>>
>> http://www.swiss.ai.mit.edu/~jaffer/r5rs_5.html#SEC20
>>
>> So we'd have to caveat the language standard, and surprise newcomers 
>> to JScheme from other Schemes.  On the principle of least surprise, 
>> we'd probably be 50-50 between surprising newcomes to JScheme from 
>> other languages vs. other Schemes.
>>
>> In Java you can't do:  Object o = NULL; if (o) { true-stuff} else { 
>> false-stuff }
>> you have to do:  if (o == NULL) ...
>> It's a pain in the butt, but it's been drilled into us.
>>
>> Reluctantly, I suggest leaving #null and the semantics of `if' alone, 
>> and offering a "new" `if', e.g., `jif', that would treat #null as #f.
>>
>> By the way, I tried to figure out what #null is in JScheme, and it 
>> isn't anything:
>>
>> > (symbol? #null)
>> #f
>> > (boolean? #null)
>> #f
>> > (pair? #null)
>> #f
>> > (vector? #null)
>> #f
>> > (procedure? #null)
>> #f
>> > (string? #null)
>> #f
>> > (port? #null)
>> (port? '#null)
>>   ====================================
>> SchemeException: ERROR: undefined variable "port?"
>> > (number? #null)
>> #f
>> > (char? #null)
>> #f
>>
>> Then I remembered that the empty list '() isn't anything either, but 
>> it also isn't #null:
>>
>> > (eq? '() #null)
>> #f
>> > (eqv? '() #null)
>> #f
>> > (equal? '() #null)
>> #f
>
> and, of course,
>
> > (null? #null)
> #f
>
> As an alternative, what about defining #f and #null as the same object?
>
>
>
>
> -------------------------------------------------------
> This SF.Net email is sponsored by:
> Sybase ASE Linux Express Edition - download now for FREE
> LinuxWorld Reader's Choice Award Winner for best database on Linux.
> http://ads.osdn.com/?ad_id=5588&alloc_id=12065&op=click
> _______________________________________________
> Jscheme-user mailing list
> Jsc...@li...
> https://lists.sourceforge.net/lists/listinfo/jscheme-user

Re: [Jscheme-user] treating #null as #f in conditionals?

[Alan D. <ado...@cs...>]

On Thu, Nov 04, 2004 at 03:55:33PM -0500, Timothy John Hickey wrote:
> One problem with identifying #f and #null is that they have
> different semantics in Java and hence would create problems when
> calling Java methods/constructors etc.

> e.g. to create a Boolean array of size 4 and store two Boolean values 
> in it, we could use
> 
>   >   (define f (list->array Boolean.class  '( #f #t #null #null)))
> 
> But if #f and #null were identified, we would not be able to represent 
> #null
> 

I agree; I think it's much cleaner to keep #null distinct from #f.

I used to think that 'null' was simply the "absence of objects", but I
have recently come to think of it as an object in its own right,
albeit a special one on which method and field accesses all throw
exceptions, and which is a subclass of everything, an "instanceof" [*]
nothing, and is equal only to itself.

Here's a somewhat tangential explanation for that statement.  I'm
working on pointer analysis: it abstracts the value of an expression
by the set of object allocation sites it can point to.  Commonly, one
represents the null pointer by the empty set; however, this is not
always a good choice.  One application of pointer analysis is to prove
that an expression cannot point to 'null', so that a region of code
can be proven free of null-pointer exceptions (so one can optimise
away dynamic null checks).  In such an analysis, it is essential to
represent the null pointer as an explicit object; then, an empty
points-to set means that an expression has no value at all, not that
it is (or may be) null.

In this view of the world, #null should satisfy exactly the same set
of string?/list?/null? predicates as does any other instance of
java.lang.Object (which according to R4RS, is none, I suppose).

alan




[*] for "instanceof", read "non-null instance of a subclass of".

Re: [Jscheme-user] treating #null as #f in conditionals?

[Ken A. <kan...@bb...>]

Thanks for your description of #null.  Its interesting that Scheme, and functional languages can get along without it.  Why is that?  I think Common Lisp had a #null called NIL.

My plan was that #null and #f would be distinct objects, but that #null would be treated as #f.  in conditionals.

The problem with Rusty and Geof's macros is you need to provide ifj, andj, orj, notj, and condj.  Here's a piece of code that's written in this style:

       (let ((effective (.getNotamEffectiveDtg n))
             (expire (.getNotamExpireDtg n)))
         (if (andj effective expire)
             (prefix " WEF " (string-append (.substring effective 2) "-"
                                            (.substring expire 2)))
             (ifj expire (prefix " TIL " (.substring expire 2))
                  (ifj effective (prefix " WEF " (.substring effective 2)))))))

which i don't like.

treating #null like #f in conditions lets  java collections return #null but let you write code like (if (member x y) ... or (if (.get table key) ...

k
At 04:16 PM 11/4/2004 -0500, Alan Donovan wrote:
>I agree; I think it's much cleaner to keep #null distinct from #f. 

Re: [Jscheme-user] treating #null as #f in conditionals?

[Timothy J. H. <ti...@cs...>]

On Nov 4, 2004, at 5:44 PM, Ken Anderson wrote:

> Thanks for your description of #null.  Its interesting that Scheme, 
> and functional languages can get along without it.  Why is that?  I 
> think Common Lisp had a #null called NIL.
>
> My plan was that #null and #f would be distinct objects, but that 
> #null would be treated as #f.  in conditionals.
That seems more reasonable as it just extends the semantics of some 
primitives and special forms but doesn't change
the data model of the language.
>
> The problem with Rusty and Geof's macros is you need to provide ifj, 
> andj, orj, notj, and condj.  Here's a piece of code that's written in 
> this style:
>
>        (let ((effective (.getNotamEffectiveDtg n))
>              (expire (.getNotamExpireDtg n)))
>          (if (andj effective expire)
>              (prefix " WEF " (string-append (.substring effective 2) 
> "-"
>                                             (.substring expire 2)))
>              (ifj expire (prefix " TIL " (.substring expire 2))
>                   (ifj effective (prefix " WEF " (.substring effective 
> 2)))))))
>
> which i don't like.
I agree, combining if and ifj, or and orj, etc gets too messy and 
creates new sources of error.
.
The alternative of changing the standard definition of if, cond, or, 
and, etc to treate #null as #f
seems reasonable to me. If someone didn't like it they could always use 
a module that
redefines them to their original meanings...

---Tim---

>
> treating #null like #f in conditions lets  java collections return 
> #null but let you write code like (if (member x y) ... or (if (.get 
> table key) ...
>
> k
> At 04:16 PM 11/4/2004 -0500, Alan Donovan wrote:
>> I agree; I think it's much cleaner to keep #null distinct from #f.
>
>
>
> -------------------------------------------------------
> This SF.Net email is sponsored by:
> Sybase ASE Linux Express Edition - download now for FREE
> LinuxWorld Reader's Choice Award Winner for best database on Linux.
> http://ads.osdn.com/?ad_id=5588&alloc_id=12065&op=click
> _______________________________________________
> Jscheme-user mailing list
> Jsc...@li...
> https://lists.sourceforge.net/lists/listinfo/jscheme-user

Re: [Jscheme-user] treating #null as #f in conditionals?

[Ken A. <kan...@bb...>]

Not quite, (or) is built into the evaluator.

At 06:11 PM 11/4/2004 -0500, Timothy John Hickey wrote:
>The alternative of changing the standard definition of if, cond, or, and, etc to treate #null as #f
>seems reasonable to me. If someone didn't like it they could always use a module that
>redefines them to their original meanings... 

Re: [Jscheme-user] treating #null as #f in conditionals?

[Ken A. <kan...@bb...>]

This is a good point.  This is why i wanted language lawyers/developers to weigh in.

JScheme has taken the policy of "bending the rules toward Java".  My question is, really, is this change worth it from the Java/Scheme perspective and our current code base.

k

At 05:59 PM 11/3/2004 -0500, Geoffrey Knauth wrote:
>As a practical matter, it certainly seems convenient to be able to treat #null the same as #f in conditionals--I agree it is counterintuitive to have #null be not-false.
>
>But I'm concerned about breaking R5RS 6.3.1., "only #f counts as false in conditional expressions,"
>
>http://www.swiss.ai.mit.edu/~jaffer/r5rs_8.html#SEC58
>
>note:  (not '())  ==>  #f
>
>and 3.2, "no object satisfies more than one of the following predicates: boolean? pair? symbol? number? char? string? vector? port? procedure?"
>
>http://www.swiss.ai.mit.edu/~jaffer/r5rs_5.html#SEC20
>
>So we'd have to caveat the language standard, and surprise newcomers to JScheme from other Schemes.  On the principle of least surprise, we'd probably be 50-50 between surprising newcomes to JScheme from other languages vs. other Schemes.
>
>In Java you can't do:  Object o = NULL; if (o) { true-stuff} else { false-stuff }
>you have to do:  if (o == NULL) ...
>It's a pain in the butt, but it's been drilled into us.
>
>Reluctantly, I suggest leaving #null and the semantics of `if' alone, and offering a "new" `if', e.g., `jif', that would treat #null as #f.
>
>By the way, I tried to figure out what #null is in JScheme, and it isn't anything:
>
>> (symbol? #null)
>#f
>> (boolean? #null)
>#f
>> (pair? #null)
>#f
>> (vector? #null)
>#f
>> (procedure? #null)
>#f
>> (string? #null)
>#f
>> (port? #null)
>(port? '#null)
>  ====================================
>SchemeException: ERROR: undefined variable "port?"
>> (number? #null)
>#f
>> (char? #null)
>#f
>
>Then I remembered that the empty list '() isn't anything either, but it also isn't #null:
>
>> (eq? '() #null)
>#f
>> (eqv? '() #null)
>#f
>> (equal? '() #null)
>#f
>
>Geoffrey
>--
>Geoffrey S. Knauth | http://knauth.org/gsk
>
>On Nov 3, 2004, at 16:32, Ken Anderson wrote:
>
>>Here's a proposal.
>>
>>I think (and i know Rusty really does) think JScheme handling of #null is awk weird.  What if in conditionals we treated #null as #f.  This would make code more scheme like, here are 2 examples.
>>
>>#null is currently treated at #t in conditionals but this change may not break much code.
>>
>>
>>(define (default  value default)
>>  ;; Before
>>  (if (isNull value) default
>>      value))
>>
>>(define (default value default)
>>  ;; After.
>>  (or value default))
>>
>>(define (memoize key table computation)
>>  ;; Before.
>>  (let ((it (.get key table)))
>>    (if (not (isNull it)) it
>>        (let ((it (computation key)))
>>          (.put table key it)
>>          it))))
>>
>>(define (memoize key table computation)
>>  ;; After.
>>  (or (.get key table)
>>      (let ((it (computation key)))
>>          (.put table key it)
>>          it)))

Re: [Jscheme-user] treating #null as #f in conditionals?

[Robert J. B. <ru...@bb...>]

Ken,
While you reference my preferences in your initial note, I think that 
Geoffrey has a good point. I guess most of my (and your) use cases would 
be easily satisfied with a macro that was a short-circuit operation like 
or (didn't evaluate the second or later arguments if it didn't need to) 
and treated #null as or treats #f. Thus (orJ x y) would return x if x != 
#null, and y otherwise. This doesn't simplify (if x y z), but Geoffrey's 
jif would work there...

--Rusty


Ken Anderson wrote:

>This is a good point.  This is why i wanted language lawyers/developers to weigh in.
>
>JScheme has taken the policy of "bending the rules toward Java".  My question is, really, is this change worth it from the Java/Scheme perspective and our current code base.
>
>k
>
>At 05:59 PM 11/3/2004 -0500, Geoffrey Knauth wrote:
>  
>
>>As a practical matter, it certainly seems convenient to be able to treat #null the same as #f in conditionals--I agree it is counterintuitive to have #null be not-false.
>>
>>But I'm concerned about breaking R5RS 6.3.1., "only #f counts as false in conditional expressions,"
>>
>>http://www.swiss.ai.mit.edu/~jaffer/r5rs_8.html#SEC58
>>
>>note:  (not '())  ==>  #f
>>
>>and 3.2, "no object satisfies more than one of the following predicates: boolean? pair? symbol? number? char? string? vector? port? procedure?"
>>
>>http://www.swiss.ai.mit.edu/~jaffer/r5rs_5.html#SEC20
>>
>>So we'd have to caveat the language standard, and surprise newcomers to JScheme from other Schemes.  On the principle of least surprise, we'd probably be 50-50 between surprising newcomes to JScheme from other languages vs. other Schemes.
>>
>>In Java you can't do:  Object o = NULL; if (o) { true-stuff} else { false-stuff }
>>you have to do:  if (o == NULL) ...
>>It's a pain in the butt, but it's been drilled into us.
>>
>>Reluctantly, I suggest leaving #null and the semantics of `if' alone, and offering a "new" `if', e.g., `jif', that would treat #null as #f.
>>
>>By the way, I tried to figure out what #null is in JScheme, and it isn't anything:
>>
>>    
>>
>>>(symbol? #null)
>>>      
>>>
>>#f
>>    
>>
>>>(boolean? #null)
>>>      
>>>
>>#f
>>    
>>
>>>(pair? #null)
>>>      
>>>
>>#f
>>    
>>
>>>(vector? #null)
>>>      
>>>
>>#f
>>    
>>
>>>(procedure? #null)
>>>      
>>>
>>#f
>>    
>>
>>>(string? #null)
>>>      
>>>
>>#f
>>    
>>
>>>(port? #null)
>>>      
>>>
>>(port? '#null)
>> ====================================
>>SchemeException: ERROR: undefined variable "port?"
>>    
>>
>>>(number? #null)
>>>      
>>>
>>#f
>>    
>>
>>>(char? #null)
>>>      
>>>
>>#f
>>
>>Then I remembered that the empty list '() isn't anything either, but it also isn't #null:
>>
>>    
>>
>>>(eq? '() #null)
>>>      
>>>
>>#f
>>    
>>
>>>(eqv? '() #null)
>>>      
>>>
>>#f
>>    
>>
>>>(equal? '() #null)
>>>      
>>>
>>#f
>>
>>Geoffrey
>>--
>>Geoffrey S. Knauth | http://knauth.org/gsk
>>
>>On Nov 3, 2004, at 16:32, Ken Anderson wrote:
>>
>>    
>>
>>>Here's a proposal.
>>>
>>>I think (and i know Rusty really does) think JScheme handling of #null is awk weird.  What if in conditionals we treated #null as #f.  This would make code more scheme like, here are 2 examples.
>>>
>>>#null is currently treated at #t in conditionals but this change may not break much code.
>>>
>>>
>>>(define (default  value default)
>>> ;; Before
>>> (if (isNull value) default
>>>     value))
>>>
>>>(define (default value default)
>>> ;; After.
>>> (or value default))
>>>
>>>(define (memoize key table computation)
>>> ;; Before.
>>> (let ((it (.get key table)))
>>>   (if (not (isNull it)) it
>>>       (let ((it (computation key)))
>>>         (.put table key it)
>>>         it))))
>>>
>>>(define (memoize key table computation)
>>> ;; After.
>>> (or (.get key table)
>>>     (let ((it (computation key)))
>>>         (.put table key it)
>>>         it)))
>>>      
>>>
>
>
>
>-------------------------------------------------------
>This SF.Net email is sponsored by:
>Sybase ASE Linux Express Edition - download now for FREE
>LinuxWorld Reader's Choice Award Winner for best database on Linux.
>http://ads.osdn.com/?ad_id=5588&alloc_id=12065&op=click
>_______________________________________________
>Jscheme-user mailing list
>Jsc...@li...
>https://lists.sourceforge.net/lists/listinfo/jscheme-user
>
>
>  
>