[CM] linux, sbcl, osc patch

Ralf Mattes rm at seid-online.de
Tue, 04 Dec 2007 15:38:28 +0100


On Tue, 2007-12-04 at 06:15 -0600, Rick Taube wrote:
> >
> > Well, I don't know how asdf works (AFAIK you started using it in cm  
> > - I
> > then tried to replicate that for clm and cmn and it was a pain to get
> > everything packaged).
> >
> > Furthermore, you are using "use-system" to load systems, which is a
> > function supplied by Common Music itself, with its own logic.
> 
> nando its got nothing to do with cm or with use-system -- its ASDF  
> thats having a problem. even if i zero out *central-registry* on  
> ccrma it still looks under sbcl.
> 
> * (setq asdf::*central-registry* ())

Ah, but that affects only systems that haven't already been "found" by
asdf. What does the following code produce when you run it just after
you started your Lisp?

  (loop 
     for key being each hash-key of asdf::*defined-systems* using
(hash-value system)
     do (format t "~%Found system ~s in ~s" key
(asdf::component-relative-pathname (cdr system))))

> * (progn
>         (trace load)
>         (pushnew "/user/h/hkt/cffi-070221/"  asdf::*central-registry*)
>         (asdf::oos 'asdf:load-op 'cffi)
>         )


It pretty much seems as if the cffi system was already found by asdf -
altering the central registy will only affect systems not yet found.

> 
> this sounds to me like a probelm with sbcl's support of asdf. asdf is  
> working in CLISP on CRRMA:

Most likely because CLISP dosen't come with a pre-defined place where
asdf looks for systems. Just to make shure you aren't hunting ghosts:
your SBCL image does not contain any traces of asdf system objects?


 HTH Ralf Mattes