Re: file-readable-p returns t for empty string

[David Kastrup]

rms@gnu.org (Richard Stallman) writes:

>     "Absolute file name" is a widely-used notion, and as such, is mostly 
>     known to programmers, even if its exact definition is sometimes a bit 
>     vague.  An empty string is a strange file name to include in that 
>     category, so programmers will probably not expect that.
> 
> Yes, exactly.  The idea of defining "" as an absolute file name is
> absurd, so there is no need to investigate whether it would "work".
> We won't do this.

Well, expanding "" (which would yield an error if you tried it opening
under any operating system) to ".", an entirely different file, is
equally absurd.  While there is a perverse logic to calling ""
absolute (as it is supposed to fail regardless of the current working
directory setting), I agree that the logic is, well, perverse.

I think it would be more to the expectations of the usual programmer
if
a) (expand-file-name "") returned ""
and
b) (file-name-absolute-p "") returned NIL

After all, the file name is illegal.  Both declaring it as equivalent
to the legal file name "." as well as declaring it absolute seem
absurd.


-- 
David Kastrup, Kriemhildstr. 15, 44793 Bochum
Email: David.Kastrup@t-online.de