Re: wild(carded)-modules for asdf

Rahul Jain <[email protected]> writes:

> I've implemented modules which take their components from a
> wild-pathname, seemingly. It was quite simple (once I passed a parent
> keyword arg to make-instance for the sub-components...) to
> implement. Good work, Dan. :)

Cool stuff.  I'm looking at merging this into asdf.lisp itself (where
it will need to be if end-users are going to be able to use it, as
asdf.lisp is supposed to be the only file they need).  Couple of
changes I plan to make

1) Remove your component-type from wild-module and add
   overrride-component-type to module.  Given that we have the
   functionality I can't see a real excuse for only having here
   instead of also there

2) The problem with creating components at reinitialize-instance time
   is that it won't pick up on files added/deleted after the system
   file is loaded.  So, I propose to have that stuff happen on a call
   to module-components instead.  It should cache the answers (a) to
   avoid repeated calls to DIRECTORY, which at least in SBCL will
   expand to a truly absurd number of stat() syscalls, and (b) to
   ensure that objecte identity is preserved at least to the extent
   that traversal and dependency checking functions

I have the basis of a patch that will do the trick, but this margin is
too small to contain it.


-dan

-- 

  http://ww.telent.net/cliki/ - Link farm for free CL-on-Unix resources 

_______________________________________________________________

Don't miss the 2002 Sprint PCS Application Developer's Conference
August 25-28 in Las Vegas - http://devcon.sprintpcs.com/adp/index.cfm?source=osdntextlink