Re: [Sbcl-devel] subtype referencing EQL or MEMBER

[Douglas K. <do...@go...>]

Jean-Philippe Paradis, You're my hero.

This is compelling evidence in favor of half of
https://bugs.launchpad.net/sbcl/+bug/922952 which I filed saying that the
compiler should be free to transform (KEYWORDP :ANY-KEYWORD) into the
constant T.
(The negative case is not true because of point 1).

On Wed, Jan 13, 2016 at 4:19 PM, Jean-Philippe Paradis <hex...@gm...>
wrote:

> Uninterning of keywords has undefined consequences, since:
>
> 1. CLHS Type KEYWORD (
> http://www.lispworks.com/documentation/HyperSpec/Body/t_kwd.htm)
> specifies:
>
> "Interning a symbol in the KEYWORD package has three automatic effects:
> [...] 2. It causes the symbol to become an external symbol of the KEYWORD
> package."
>
> 2. CLHS Function UNINTERN (
> http://www.lispworks.com/documentation/HyperSpec/Body/f_uninte.htm)
> specifies:
>
> "If package is the home package for symbol, symbol is made to have no home
> package."
>
> 3. CLHS System Class SYMBOL (
> http://www.lispworks.com/documentation/HyperSpec/Body/t_symbol.htm)
> specifies:
>
> "The consequences are undefined if an attempt is made to alter the home
> package of a symbol external in the COMMON-LISP package or the KEYWORD
> package." (Easily found by searching for "keyword" at
> http://www.hexstreamsoft.com/articles/notes-tips-standard-common-lisp-symbols/themes/unspecified-behavior/,
> btw.)
>
>
>
>> On Wed, Jan 13, 2016 at 2:26 PM, Douglas Katzman <do...@go...>
>> wrote:
>>
>>> You could file a bug. Some folks may point out that keywordness of
>>> symbols can be changed anyway and that the subtypep test is not
>>> well-defined:
>>>
>>> * (defvar *x* :x)
>>> *X*
>>> * (typep *x* 'keyword)
>>> T
>>> * (unintern *x* 'keyword)
>>> T
>>> * (typep *x* 'keyword)
>>> NIL
>>>
>> http://pubads.g.doubleclick.net/gampad/clk?id=267308311&iu=/4140
>> _______________________________________________
>> Sbcl-devel mailing list
>> Sbc...@li...
>> https://lists.sourceforge.net/lists/listinfo/sbcl-devel
>>
>>
>