Re: Is this correct?

Hi Tim,

I dabbled exactly in that many years ago; my NAPS library is still floating
around.

The CL pathname spec is not *that* bad.  What is bad is that pathname-device
and pathname-host etc return different values in the current different
implementations, making things like UIOP both necessary and difficult to
implement as Farè teaches us.

TRT to do would be to have an addendum on the ANSI spec that said things
like the following (YMMV).

   - *On Windows* pathname-device returns one of the drive letters (Martin
   loves this, does he not? 😏 )
   - *On Windows* pathname-type cannot contain dots.
   - *On Unix *pathname-version consistently returns NIL (or :unspecific,
   as long as it is one and only one of the two)
   - *On Unix* truename returns the "name" associated with the first
   created instance of the file (IIRC you can get that from the inode; correct
   me if I am wrong - BTW: Allegro Express works exactly like SBCL).
   - *On Unix* pathname-type cannot contain dots.
   - *On z/OS* truename returns the member from the catalogued PDS on the
   original volume (whatever that means)
   - ect ect etc (*)

Note that I do not care, e.g., whether the pathname type contains dots or
not.  What I care about is that all implementations behave the same.

But, as usual, we are opening up the "we need an addendum to the ANSI" can
of worms.  So you guys will just have to bear with me from time to time.
Sorry. 🥹

Cheers

MA

(*) As you see I mentioned Unix, Windows and z/OS. I could mention z/VM,
VMS or MTS.  I will *not* mention ITS, TENEX or TOPS-20 etc. for the well
known reasons... 🙄🥹🫩






On Sat, Feb 21, 2026 at 6:32 PM Tim Bradshaw <[email protected]> wrote:

> On 20 Feb 2026, at 18:01, Marco Antoniotti <[email protected]>
> wrote:
> >
> > I think it was Christopher Stacy who just said that the ANSI committee
> "punted" on pathnames.
>
> And I will repeat what I have said before: people who complain about the
> CL pathname system (or more generally about design decisions in any
> language) without coming up with a detailed proposed replacement design
> with at least one implementation are acting like fools.
>
> Especially as here where it's pretty clear that LW's truename
> implementation just ought to return, well, 'the canonical filename of the
> file in the filesystem' as the spec requires and it's really not doing that.
>
> --tim



-- 
Marco Antoniotti, Professor, Director         tel. +39 - 02 64 48 79 01
DISCo, University of Milan-Bicocca U14 2043   http://dcb.disco.unimib.it
Viale Sarca 336
I-20126 Milan (MI) ITALY

REGAINS: https://regains.disco.unimib.it/