pathnames [Re: files, printers, etc. [Re: The feel of a Lisp

Chris Bitmead uid(x22068)
Tue, 06 May 1997 11:34:29 +1000

>> I think this is subtley locking yourself into a particular paradigm of
>> a Unix style tree structure.
>Not necessarily - the only thing I am asserting here is that groups
>of objects may contain groups of objects. Those symbols, that
>reference objects inside groups, could be anything - a group of
>large-number prime factorisations, 

Well groups within groups is a tree structure by definition.

>say, could use numeric
>identifiers, and an analog associative store might accept
>pictures as identifiers, returning the object most closely
>matching - ie, there is an n:1 mapping of identifiers to objects.

Yes, so if you want to use pictures as identifiers that implys that
there could be several algorithms for getting at the object you
want. It might be you want to get objects by name, by colour, by
description, by location etc. So a simple (foo bar baz) meaning "the
baz object in the bar collection in the foo collection" is not enough

>> >((.) Hello!)
>> If you want a CWD you can just store one in a variable.
>That's what the . was, rather than a global concept... just a standard

Well if it's a standard then that kind of implys a global
concept. Perhaps I don't understand you right.

>The CWD is really only a user-level concept anyway; it only complicates
>programming tasks IMHO.
>=> a new syntax:
>(make-path *internet* "" 'Public 'Hello!)

You're still assuming simple tree structure with a single key for
getting at your objects.