Re: bugs in dirname module

[Eric Blake]

> > Actually, I think that calling base_name should strip the trailing
> > slashes, similar to POSIX basename().
> 
> But if you do that, you violate this basic rule:
> 
> >   If you have a valid file name F, then accessing F is the same as
> >   chdir(dir_name(F)) followed by accessing base_name(F).  Furthermore
> >   if you successfully { chdir(dir_name(F)); rename(base_name(F),"foo"); },
> >   you have renamed F to a file named "foo" in the same directory that F was 
> in.
> If a/b is a symbolic link to a directory, then a/b/ is not the same as
> a/b and should not be treated like it.  Hence base_name shouldn't
> strip trailing slashes.

OK, I missed the finer points of that on my first read.  My patch
will have the malloc'ing version (whether that will be spelled
base_name or something else still remains to be seen) preserve
the trailing slashes, so that rename(base_name("symlink","foo"))
behaves differently than rename(base_name("symlink/","foo")),
along with a comment that for POSIX basename() semantics, you
must use strip_trailing_slashes(base_name()).

I guess that also means that base_len() is still useful;
base_len(base_name("symlink/")) is smaller than
strlen(base_name("symlink/")), without modifying the
resulting string.

--
Eric Blake