Re: [Sbcl-devel] declare and shadowing.

[Peter S. <pe...@pj...>]

By picking an exact symbol to use for the placeholder (instead of 
any-symbol-with-name), the standard lets code that processes array type 
declarations (mostly the compiler, but code-walkers and macros get 
involved here too) use EQ (fast) instead of 
some-test-ultimately-involving-STRING= (slow). This comes at the minor 
cost that people doing strange things (shadowing CL:* hint, hint) will 
be inconvenienced (but not inhibited, since you can access the correct 
symbol using the package syntax). I don't think there is anywhere in the 
standard that requires the use of a symbol-of-a-particular-name without 
also specifying a package (generally the package CL)

For some problems like this you could do the equivalent of shadowing 
DECLARE with your own TEST-DECLARE2:DECLARE which would do the 
appropriate mapping so you didn't have to worry about using the package 
prefix for *, but (again for efficiency reasons) you are explicitly 
denied writing macros that expand into (pure) DECLARE forms.

On 21/05/15 01:42, Harvey Stein wrote:
> I'm surprised by the behavior of declare when * is shadowed.  In 
> particular, the following works fine:
>
> (defpackage :test-declare1
>   (:use :common-lisp)
>   (:shadow "*"))
>
> (in-package :test-declare1)
>
> (defun fcn (x)
>   (declare (type (simple-array double-float (cl:*)) x))
>   (cl:* (aref x 1) 2d0))
>
>
> but, the following signals an error:
>
> (in-package :common-lisp)
>
> (defpackage :test-declare2
>   (:use :common-lisp)
>   (:shadow "*"))
>
> (in-package :test-declare2)
>
> (defun fcn (x)
>   (declare (type (simple-array double-float (*)) x))
>   (cl:* (aref x 1) 2d0))
>
>
> ; file: /home/hjstein/lisp/cls/packtest2.lisp
> ; in: DEFUN FCN
> ;     (TYPE (SIMPLE-ARRAY DOUBLE-FLOAT (TEST-DECLARE2::*)) 
> TEST-DECLARE2::X)
> ;
> ; caught ERROR:
> ;   bad dimension in array type: *
> ;
> ; compilation unit finished
> ;   caught 1 ERROR condition
>
> I was surprised because semantically, the "*" in the (declare ...) 
> form is just a place holder for the dimension.  Why does it have to be 
> the actual cl:* symbol and not just any symbol with name "*"?
>
> Thanks,
> Harvey
>
> -- 
> Harvey J. Stein
> hj...@gm... <mailto:hj...@gm...>
> http://www.linkedin.com/in/harveyjstein
> Selected papers and presentations available at: 
> http://ssrn.com/author=732372
>
>
> ------------------------------------------------------------------------------
> One dashboard for servers and applications across Physical-Virtual-Cloud
> Widest out-of-the-box monitoring support with 50+ applications
> Performance metrics, stats and reports that give you Actionable Insights
> Deep dive visibility with transaction tracing using APM Insight.
> http://ad.doubleclick.net/ddm/clk/290420510;117567292;y
>
>
> _______________________________________________
> Sbcl-devel mailing list
> Sbc...@li...
> https://lists.sourceforge.net/lists/listinfo/sbcl-devel