Re: ASDF vs. logical pathnames in CMUCL

[email protected] (Björn Lindberg) writes:

> I do the same thing, except for one difference which also solves the
> FASL issue. I have the following in my startup script for CMUCL:
[...]
> (defmethod output-files :around ((operation compile-op) (c source-file))
>   (let ((source (component-pathname c))

For what it's worth, now that CLISP supports non-default method
combinations, I /may/ define a custom method combination for 
output-files so that all the different people who might want to define
policy (the package author, the lisp implementation distributor,
#+linux (the distribution vendor), the local sysadmin, the end-user) 
get to have a say without all having to be concerned whether someone
else is using the :around method.  Needs a bit of design first, though.

> I don't see that this couldn't work equally well with logical
> pathnames in *system-configuration-paths*, as long as ASDF in general
> supports them.

The problem Christopher Stacy was experiencing seemed to be due to a
slightly dubious lack-of-case-folding in CMUCL, for which Raymond Toy
has since committed a CMUCL fix.  (Thanks, Ray, if you're reading this)

Some form of workaround in ASDF until the 19a CMUCL release is
available and people are using it might nevertheless be appropriate.


-dan

-- 
"please make sure that the person is your friend before you confirm"

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.2.4 (GNU/Linux)

iD8DBQBAqjlPHDK5ZnWQiRMRAsAlAJ0Sa6w/BgEWcuZrKX9JqD+fRTjlPACfb1us
8gsx2g+1Ps1dW7f2JCGNQa4=
=ojoQ
-----END PGP SIGNATURE-----