From: jqb AT netcom DOT com (Jim Balter) Subject: Re: Minor problem with du output in beta 17.1 4 Apr 1997 22:40:25 -0800 Approved: cygnus DOT gnu-win32 AT cygnus DOT com Distribution: cygnus Message-ID: <334592B4.3CD6.cygnus.gnu-win32@netcom.com> References: <13409 DOT 9704040901 AT egor DOT pdd DOT 3com DOT com> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit X-Mailer: Mozilla 3.01Gold (WinNT; I) Original-To: Ian Macpherson Original-CC: gnu-win32 AT cygnus DOT com Original-Sender: owner-gnu-win32 AT cygnus DOT com Ian Macpherson wrote: > I believe the output of the gnuish 'du' to be correct. Compared to this the > gnu-win32 'du' misses sub-directories and seems to be reporting 2K blocks > (even if -k is used). du under gnu-win32 is known to report in multiples of 2048 rather than 1024 because stat returns the wrong value for st_blocks; try du -b or du -h instead (-k is the default unless POSIXLY_CORRECT is set in the environment). It is also known to ignore some files or directories altogether because it uses inode numbers to check for loops but stat returns non-unique inode numbers. You could patch the du source to not look for loops. > As a general point, I find the gnu-win32 utilities very slow when comapred > to the gnuish DOS equivalents, particularly programs that traverse directories. > Is there a reason for this? gnu-win32 stat opens every file to try to see if it is a symlink or executable. Maybe it will come to its senses in b18. -- - For help on using this list, send a message to "gnu-win32-request AT cygnus DOT com" with one line of text: "help".