[Openmcl-devel] when is :<BOOL> == T?

alex crain alexcrain at mail.widgetworks.com
Fri Aug 5 17:07:40 UTC 2005



I was working with this method definition

(define-objc-method ((:unsigned :dragging-source-operation-mask-for- 
local (:<BOOL> is-local))
              hemlock-text-view)
   (if (= is-local #$YES)
       (logior #$NSDragOperationCopy #$NSDragOperationMove)
     #$NSDragOperationCopy))

Which throws an exception because is-local is either T or nil, not # 
$YES or #$NO.

This kind of makes sense and kind of doesn't

What are the rules for :<BOOL> objects? are they always translated  
correctly, so that

(send view :set-hidden T)
is correct and
(send view :is-hidden) => T
?
Can I/Should I ever do (send view :set-hidden #$YES)

It would seem like if this were going to be the case, then #$YES  
should EQ T and #$NO should EQ NIL.

This gets more complex because if I don't specify the argument type,  
as in

(define-objc-method ((:unsigned :dragging-source-operation-mask-for- 
local is-local)
              hemlock-text-view)
   (if  (= is-local #$YES)
       (logior #$NSDragOperationCopy #$NSDragOperationMove)
     #$NSDragOperationCopy))

is-local now comes through as #$YES or #$NO.

The whole thing strikes me as reverse-intuitive: if I know enough to  
specify the ObjC type for a variable, OpenMCL
converts it to an incompatible lisp type. If I don't know what it is,  
it comes over as the raw ObjC value.

I like the idea of making ObjC booleans into simple lisp booleans,  
but it seems more confusing then it's worth.
Maybe it would be better to just have an OBJC-IS-TRUE primitive?

:alex





More information about the Openmcl-devel mailing list