Re: [Kawa-commonlisp-dev] [GSoC] Status

[Jamison Hope]

Hi, Charlie,

On Aug 9, 2012, at 1:58 PM, Charles Turner wrote:

> I've cleaned up the repository, it should build without any problems  
> now. I've tried to reduce the problem by commenting out most of the  
> new def*Fld's in CommonLisp, this is what I've noticed:
>
> * With all the def*Fld's commented out, I would get warnings when  
> compiling the primitives.lisp files about LENGTH, GENSYM and SYMBOL- 
> NAME being unbound, this made sense because their call sites were  
> before their definitions (and hence the warnings would disappear  
> compiling again, if loadClass(primitives) is not commented out).  
> Adding def*Fld's for these three functions caused an error rather  
> than a warning when building the primitives from scratch:
>
> primitives.lisp:152:23: exception loading 'COMMON-LISP:SYMBOL-NAME'  
> - no such class: gnu.commonlisp.lisp.primitives
> [ and similar messages for the other two examples ]
>
> * Once the primitives file is compiled and I use loadClass on it,  
> everything appears to work fine, the names are bound from the REPL,  
> etc.
>
> * If you remove the loadClass("gnu.commonlisp.lisp.primitives") line  
> after the primitives file has been compiled, no names appear to be  
> bound from the REPL, for example, assuming this is uncommented in  
> CommonLisp:
>
> defProcStFld("gensym", "gnu.commonlisp.lisp.primitives");
>
> Then from the REPL I see:
>
> #|kawa:1|# #'gensym
> /dev/stdin:1:3: warning - no declaration seen for COMMON-LISP:GENSYM
> /dev/stdin:1:3: unbound location
>    at gnu.kawa.reflect.FieldLocation.get(FieldLocation.java:200)
>    at gnu.kawa.reflect.FieldLocation.get(FieldLocation.java:188)
>    at gnu.mapping.PlainLocation.get(PlainLocation.java:21)
>    at gnu.mapping.SharedLocation.get(SharedLocation.java:21)
>    [ snipped ... ]
>
> which is the same output as when the above defProcStFld(gensym..) is  
> commented out, so it appears to do nothing at the moment.

I'm seeing that as well. I moved LENGTH, GENSYM, and SYMBOL-NAME up in
the file a bit so that their definitions come before their first calls
and that got rid of the "exception loading 'COMMON-LISP:LENGTH' - no
such class: gnu.commonlisp.lisp.primitives" and "exception loading
'COMMON-LISP:LENGTH' - null" errors.

With the loadClass() call in CommonLisp commented out, the functions  
remain
unbound until gnu.commonlisp.lisp.primitives is explicitly loaded:

$ kawa --lisp
#|kawa:1|# #'gensym
/dev/stdin:1:3: warning - no declaration seen for COMMON-LISP:GENSYM
/dev/stdin:1:3: unbound location
        at gnu.kawa.reflect.FieldLocation.get(FieldLocation.java:200)
        at gnu.kawa.reflect.FieldLocation.get(FieldLocation.java:188)
        at gnu.mapping.PlainLocation.get(PlainLocation.java:21)
        at gnu.mapping.SharedLocation.get(SharedLocation.java:21)
        at gnu.mapping.ThreadLocation.get(ThreadLocation.java:105)
        at atInteractiveLevel$1.run(stdin:1)
        at gnu.expr.ModuleExp.evalModule2(ModuleExp.java:309)
        at gnu.expr.ModuleExp.evalModule(ModuleExp.java:210)
        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:2|# (require class::|gnu.commonlisp.lisp.primitives|)
#|kawa:3|# #'gensym
#<procedure GENSYM>


So for now I guess leave it in there. Not sure why it works for Scheme
but not for Lisp. There's clearly some piece of the magic that we're
missing.

Jamie

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