Re: If it looks like a bug and smells like a bug .....

No, they do not appear before &KEY, they appear right where they should.

The lambda list of SOME-FCN is (record-type &rest initargs &key foo). 
Your call looks like (some-fcn 'record :x 1 :y 2). Variable RECORD-TYPE 
is bound to the symbol RECORD. INITARGS is bound to the list (:X 1 :Y 
2). Then keyword parameter processing kicks in, performing keyword 
argument checking as per CLHS 3.5.1.4, and an error is signaled because 
:X is not listed as a valid keyword in the lambda list of SOME-FCN 
(unlike :FOO).

Everything seems OK here for me.

On 5.05.2025 16:37, Paul Werkowski wrote:
> :x & :y sould NOT be recognized as keywords because they appear before 
> &key.
>
> On 5/5/2025 10:29 AM, Michał "phoe" Herda wrote:
>> In your example, TEST1 is calling SOME-FCN with keys :X and :Y, 
>> neither of which is specified in SOME-FCN's lambda list, so an error 
>> is signaled. This is exactly the case specified in CLHS 3.5.1.4 
>> Unrecognized Keyword Arguments.
>>
>> What exactly is the bug here?
>>
>> On 5.05.2025 16:21, Paul Werkowski (as pw at snoopy dot qozzy dot 
>> com) wrote:
>>> From LispWorks CLIM 2.0 User Guide:
>>>
>>> invoke-with-output-to-output-record Generic Function
>>>
>>> (invoke-with-output-to-output-record
>>>  stream continuation record-type &rest init-args &key)
>>> ......
>>> continuation is a function of two arguments, the stream and the 
>>> output record; it has dynamic extent. record-type is the type of 
>>> output record to create. init-args are CLOS initialization arguments 
>>> that are passed to make-instance when the new output record is created.
>>>
>>> CLTL page 76 deals with parsing lambda-lists. The &rest var is a 
>>> list of everything else up to the ending #\). Keywords are not 
>>> recognized as such until after the &key marker.
>>>
>>> Consider this example:
>>>
>>> (defclass record ()
>>>   ((s1 :initarg :x)
>>>    (s2 :initarg :y)))
>>>
>>> (defun some-fcn (record-type &rest initargs &key foo)
>>>   (print initargs)
>>>   (apply 'make-instance record-type initargs))
>>>
>>> (defun test1 ()
>>>   (some-fcn  'record :x 1 :y 2))
>>>
>>> > (test1)
>>> Error: Unexpected keyword :X which is not one of (:FOO).
>>>   1 (continue) Ignore the unknown keyword :X.
>>>   2 (abort) Return to top loop level 0.
>>>
>>> This is exactly the form of error message  generated by nested 
>>> macros deep in the weeds of McCLIM table formatting code.
>>>
>>> This can be dealt with (sort of) by something like this:
>>>
>>> (defun test2 ()
>>>   (handler-bind ((conditions:unknown-keyword-error
>>>                   #'(lambda(c)
>>>                       (invoke-restart 'continue))))
>>>     (some-fcn  'record :x 1 :y 2)))
>>>
>>> > (test2)
>>> (:X 1 :Y 2)
>>> #<RECORD 401000C663>
>>>
>>>
>>> but not very useful in a large software system.
>>>
>>> Same issue with LWW 7.1.
>>>
>>> Should not be hard to fix this. Can DSPEC magic be used for a 
>>> work-around?
>>>
>>> Paul
>>>
>>>
>>>
>>>
>>> _______________________________________________
>>> Lisp Hug - the mailing list for LispWorks users
>>> [email protected]
>>> http://www.lispworks.com/support/lisp-hug.html
>

_______________________________________________
Lisp Hug - the mailing list for LispWorks users
[email protected]
http://www.lispworks.com/support/lisp-hug.html