[clisp-list] #P vs. TRUENAME consistency Was: EXT:FASTHASH-EQUAL on pathnames doesn't always return expected value...

[Pascal B. <pj...@in...>]

Sam Steingold writes:
> Pascal Bourguignon wrote:
> > P[98]> (block nil (maphash (lambda (k v) (print k) (return)) *source-=
file-db*))
> >=20
> > #P"/a6/users/pjb/src/public/lisp/clisp/susv3-xsi.lisp"=20
> > NIL
> > P[99]> (gethash #P"/a6/users/pjb/src/public/lisp/clisp/susv3-xsi.lisp=
"  *source-file-db*)
> > NIL ;
> > NIL
>=20
> pathnames are not printed readably.
> things like :version can be nil or newest without affecting the printed=
=20
> representation.
> the two identically printed pathnames above are probably different in=20
> that respect.

Ok. In the hashtable I had paths gone thru TRUENAME.

[ And I misread EQUAL, thinking that
    (=3D=3D> (functionnaly-equivalent-pathname-p a b) (EQUAL a b))
  when it's actually
    (=3D=3D> (EQUAL a b) (functionnaly-equivalent-pathname-p a b)) ]




Well #P is specified to be #.(parse-namestring 'path)
and parse-namestring is specified to use *default-pathname-defaults*
and I have (pathname-version *default-pathname-defaults*) --> :NEWEST
so I'd expect (pathname-verson #P"/tmp/a.c") to be :NEWEST, not NIL,
even if we're really in the implementation-defined clause of parse-namest=
ring.

It would give more consistency, without losing anything. People who
want a null version for #P"/tmp/a.c" could set it so in
*default-pathname-defaults*.

P[157]> (print-pathname *default-pathname-defaults*)
Host      : NIL
Device    : NIL
Directory : (:ABSOLUTE "a6" "users" "pjb" "src" "public" "lisp" "common-l=
isp")
Name      : NIL
Type      : NIL
Version   : :NEWEST
#P"/a6/users/pjb/src/public/lisp/common-lisp/"
P[158]> (print-pathname #P"/tmp/a.c")
Host      : NIL
Device    : NIL
Directory : (:ABSOLUTE "tmp")
Name      : "a"
Type      : "c"
Version   : NIL
#P"/tmp/a.c"
P[159]> (print-pathname (truename #P"/tmp/a.c"))
Host      : NIL
Device    : NIL
Directory : (:ABSOLUTE "tmp")
Name      : "a"
Type      : "c"
Version   : :NEWEST
#P"/tmp/a.c"
P[160]>=20



--=20
__Pascal Bourguignon__                     http://www.informatimago.com/

Nobody can fix the economy.  Nobody can be trusted with their finger
on the button.  Nobody's perfect.  VOTE FOR NOBODY.