[Openmcl-devel] Trunk and released version

[Ron Garret]

I dunno, I don't think copying the declarations is all that  
inexplicable.  After all, an implementation is allowed to assign VAR  
instead of rebinding it as CCL does, in which case you'd have no  
choice but to have the declarations apply in the scope of the result  
form.

rg

On Aug 13, 2009, at 8:56 AM, Gary Byers wrote:

> I'd call it a bug in the version of DOLIST that was in the trunk until
> about an hour ago.  (It was establishing a new binding of VAR to NIL
> around the result form, but inexplicably copying the declarations from
> the main body into the body of the LET or LET* that established that  
> binding.)
> So something like:
>
> (let* ((sum 0))
>  (dolist (i list sum)
>    (declare (fixnum i))
>    (incf sum i)))
>
> was expanding into something like:
>
> (uninteresting-loop
>  ...
>  (declare (fixnum i)) ; OK
>  ...
>  (when done (let* ((i nil))
>               (declare (fixnum i)) ; unintentional, I assume.
>                sum))
>
> which certainly deserves a runtime error at high safety, probably
> deserves a compile-time warning, and is a result of a bug in DOLIST.
>
>
>
>
> On Thu, 13 Aug 2009, Ron Garret wrote:
>
>> I'd call that a bug in the ANSI spec.  It violates the principle of  
>> least
>> surprise for no benefit whatsoever.  If X is unconditionally bound  
>> to NIL
>> then there is no point in even having it in scope when the result- 
>> form is
>> evaluated.
>> rg
>> On Aug 13, 2009, at 6:28 AM, Tim Bradshaw wrote:
>>
>>      On 13 Aug 2009, at 12:31, Arthur W Cater wrote:
>>
>>      a form like
>>   (dolist (X list-of-xs    expr-not-using-X)
>>       (declare (type mytype X))
>>       stuff-involving-X)
>>  provoking a complaint that NIL is not of type MYTYPE
>> I'm pretty sure that's not a bug.  X is defined to be bound to NIL
>> when the result-form is evaluated.  So unless NIL is an instance of
>> MYTYPE your declaration is wrong.
>> _______________________________________________
>> Openmcl-devel mailing list
>> Openmcl-devel at clozure.com
>> http://clozure.com/mailman/listinfo/openmcl-devel
>>