GSOC | Extending Common Lisp support

[Jamison Hope]

On Jun 9, 2012, at 10:55 AM, Charles Turner wrote:

> I've attached some bits and pieces related to my work so far with
> packages. It's all very raw and nonconforming in lots of different
> ways. It "works" in the trivial sense that I can create new package
> and move into them, but I haven't yet tested symbol resolution much,
> other than writing a SYMBOL-PACKAGE function, that current check the
> symbol's namesace, and if that's empty returns the value of *PACKAGE*.

Seems reasonable so far.

> One problem I'm having with thread locations is their return value of
> #!null. Is there a clean way of having it by default return nil that
> anyone knows of? Not actually sure if I should be using a thread
> location, but it might be useful to be able to specify a default
> package on the command line, like we do in the pretty printer with
> out:print-circle.

It *seems* like that's the purpose of defaultValue in
get(Object defaultValue), right? Except that the default value is
ignored:

#|kawa:1|# (define q-loc ::gnu.mapping.ThreadLocation  
(gnu.mapping.ThreadLocation "q"))
#|kawa:2|# (q-loc:get #f)
#!null
#|kawa:3|# (q-loc:get)
#!null

I dug through it a bit and saw that ThreadLocation#get(Object) defers
to the implementation in SharedLocation:

   public synchronized final Object get (Object defaultValue)
   {
     return base != null ? base.get(defaultValue)
       : value == Location.UNBOUND ? defaultValue : value;
   }

That says the default is only used if value==Location.UNBOUND. But  
guess what:
value (declared in IndirectableLocation) is initialized to null, not
Location.UNBOUND -- instantiating a ThreadLocation effectively defines  
its
value to be null.

So the only way to have our defaultValue returned is to set
value:=Location.UNBOUND first.

Looks like IndirectableLocation#undefine might do the trick. Let's see:

#|kawa:4|# (q-loc:undefine)
#|kawa:5|# (q-loc:get 't)
#!null

Hmm. Oh, right, gotta do it on the wrapped NamedLocation. (Is that a
bug or a feature?)

#|kawa:6|# ((q-loc:get-location):undefine)
#|kawa:7|# (q-loc:get 't)
t
#|kawa:8|# (q-loc:get #f)
#f
#|kawa:9|# (q-loc:get ())
()
#|kawa:10|# (q-loc:get)
unbound location q (property (dynamic))
	at gnu.mapping.Location.get(Location.java:67)
	at atInteractiveLevel$10.run(stdin:10)
	at gnu.expr.ModuleExp.evalModule2(ModuleExp.java:299)
	at gnu.expr.ModuleExp.evalModule(ModuleExp.java:200)
	at kawa.Shell.run(Shell.java:279)
	at kawa.Shell.run(Shell.java:194)
	at kawa.Shell.run(Shell.java:175)
	at kawa.repl.main(repl.java:891)
#|kawa:11|# *print-right-margin*
#!null
#|kawa:12|# (gnu.text.PrettyWriter:lineLengthLoc:get)
#!null
#|kawa:13|# (gnu.text.PrettyWriter:lineLengthLoc:set 5)
#|kawa:14|# *print-right-margin*
5
#|kawa:15|# ((gnu.text.PrettyWriter:lineLengthLoc:get- 
location):undefine)
#|kawa:16|# *print-right-margin*
<unknown>: warning - no declaration seen for *print-right-margin*
unbound location *print-right-margin*
	at gnu.mapping.Location.get(Location.java:67)
	at atInteractiveLevel$16.run(stdin)
	at gnu.expr.ModuleExp.evalModule2(ModuleExp.java:299)
	at gnu.expr.ModuleExp.evalModule(ModuleExp.java:200)
	at kawa.Shell.run(Shell.java:279)
	at kawa.Shell.run(Shell.java:194)
	at kawa.Shell.run(Shell.java:175)
	at kawa.repl.main(repl.java:891)


OK, it would appear that evaluating *print-right-margin* is
like invoking get() and not get(default). Depending upon what
you want to do with your ThreadLocation, maybe that's fine,
or maybe you should just call set(NIL) early on.


--
Jamison Hope
The PTR Group
www.theptrgroup.com