[Ecls-list] patch for namestring

[Juan Jose Garcia-Ripoll]

2006/10/8, Dustin Long <dlong at stevens.edu>:
> > In ANSI, having a type of NIL and having an empty string as type are
> > not the same thing.
>
> I see exactly what you mean. I suppose I was thinking far too
> Windows-centricly. A (pathname-type) = nil is more or less a
> type-wild-card, yes?

Type = NIL means, in principle, that the type is to be filled with the
value of *default-pathname-defaults*, but in most implementations this
value itself contains type = NIL.

If we were to be ANSI compliant, we would use type = :unspecific,
which means that there exists no type. However, no lisp that I know
uses this interpretation, hence I chose ECL to follow what is a
de-facto standard. (Note that there are important practical issues why
implementations chose NIL vs. :unspecific)

> Also, I had assumed that ECL's transparent parsing
> of pathname was acceptable, but you're saying that it is not?

There seems to be this bug you found. Also, thanks for offering your
help, but fixing this deserves also creating a number of test cases
and ensuring that other things are also working.

Juanjo

-- 
Max-Planck-Institut für Quantenoptik
Hans-Kopfermann-Str. 1, Garching, D-85748, Germany
Phone: +49 89 32905 345   Fax: +49 89 32905 336
http://www.mpq.mpg.de/Theorygroup/CIRAC/