This is the mail archive of the cygwin-developers@cygwin.com mailing list for the Cygwin project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: stat with no READ_CONTROL


On Thu, Jan 09, 2003 at 12:58:27PM +0100, Corinna Vinschen wrote:
>On Wed, Jan 08, 2003 at 11:15:16PM -0500, Pierre A. Humblet wrote:
>> A consequence of this hard nosed approach is that ls -l may not show 
>> all files. To avoid that, stat could return uid = gid = -1 and modes = 0.
>> 
>> Which is best in this case: conformance to posix or user friendliness?
>
>IMHO user friendliness.  I've checked in a patch.  When calling ls -l
>on such a file or dir, the output looks like this now:
>
>  ----------    1 65535    65535       48452 Jan  7 12:17 foo
>  d---------    2 65535    65535           0 Jan  9 12:09 foodir
>
>
>However, coincidentally I found another problem while testing that change.
>When a directory is unreadable for an account, the link count of that
>directory is set to 0 by the call to num_entries(). 
>
>That should be 2, shouldn't it?
>
>There's the comment in fhandler_disk_file::fstat_helper() which is
>fairly old AFAIR:
>
>  /* Unfortunately the count of 2 confuses `find (1)' command. So
>     let's try it with `1' as link count. */
>
>But this is only valid for remote dirs so it should be safe to return 2
>from num_entries.
>
>Chris?

Isn't that your comment, Corinna?

If you want to set it to 2, that's fine with me.

cgf


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]