Re: directory-files and Tramp

[Lars Hansen]

> I think it looks good: in the no-handler case, directory_files calls
> directory_files_internal, which in turn calls expand-file-name on the
> directory.  So expand-file-name is called anyway.
>
> But should the expand-file-name call in directory_files_internal be
> removed after making sure that all callers do the expansion?
Sounds like a good idea. I seems directory_files_internal is called by 
directory-files and
directory-files-and-attributes only. But the latter is missing a call to 
expand-file-name, so
it must have the same problem as directory-files. So i think the changes 
should be:

diff -c dired-original.c dired.c
*** dired-original.c    Tue Feb  4 15:03:12 2003
--- dired.c    Sat Feb 22 23:19:18 2003
***************
*** 153,159 ****
      Ffuncall, and cause a GC.  */
   list = encoded_directory = dirfilename = Qnil;
   GCPRO5 (match, directory, list, dirfilename, encoded_directory);
-   directory = Fexpand_file_name (directory, Qnil);
   dirfilename = Fdirectory_file_name (directory);

   if (!NILP (match))
--- 153,158 ----
***************
*** 346,351 ****
--- 345,352 ----
 {
   Lisp_Object handler;

+   directory = Fexpand_file_name (directory, Qnil);
+
   /* If the file name has special constructs in it,
      call the corresponding file handler.  */
   handler = Ffind_file_name_handler (directory, Qdirectory_files);
***************
*** 379,384 ****
--- 380,387 ----
 {
   Lisp_Object handler;

+   directory = Fexpand_file_name (directory, Qnil);
+
   /* If the file name has special constructs in it,
      call the corresponding file handler.  */
   handler = Ffind_file_name_handler (directory, 
Qdirectory_files_and_attributes);