Re: MSVC compiler support [patch 32]: ls

[Michael Goffioul]

John W. Eaton a écrit :
> I decided to make ls a script file function instead of a built-in
> function, and to let the user specify the command to use with a
> separate new function called ls_command.  For example,
>
>   old_cmd = ls_command ("/my/ls");
>
> returns the previous command in old_cmd and sets the internal variable
> to "/my/ls" and
>
>   curr_cmd = ls_command ();
>
> returns the current command to use.  The default value is set by this
> code:
>
>   if (isempty (__ls_command__))
>     ## FIXME -- ispc and isunix both return true for Cygwin.  Should they?
>     if (ispc () && ! isunix () && isempty (file_in_path (EXEC_PATH, "ls")))
>       __ls_command__ = "cmd /C dir /w";
>     else
>       __ls_command__ = "ls -C";
>     endif
>   endif
>
> Is it OK to use /w?  As I recall, that is the switch that causes DIR
> to print output in columns.  I think we should do that to match the
> way output is displayed by "ls -C".
>   

"/D" is better as it will sort items in columns and not in rows. This 
looks more like the UNIX-way.

> Is it reasonable to check EXEC_PATH to see whether we have a real ls
> command available?  That way, I won't be annoyed if I start up the
> Windows version of Octave and type "ls -l" (assuming I have ls in my
> PATH, I think this should work).
>   

It seems reasonable.

Michael.