On Wed, Apr 18, 2007 at 05:11:38AM +0200, Jim Meyering wrote: > Justin Pryzby <[EMAIL PROTECTED]> wrote: > > This bug is not fixed. The code in coreutils-6.9 fails to treat -D as > > an option independent of -P and -L. This has the effect that -D implies > > -L, and du -DL is not the same as du -LD. > > Please provide an example demonstrating this. > Or point to parts of the code and explain how that can be true. So I think we agree on the intended behavior.
> FYI the relevant parts sure seem orthogonal to me: int symlink_deref_bits = FTS_PHYSICAL; > case 'D': /* This will eventually be 'H' (-H), too. */ > symlink_deref_bits = FTS_COMFOLLOW | FTS_PHYSICAL; > break; > > case 'L': /* --dereference */ > symlink_deref_bits = FTS_LOGICAL; > break; > > case 'P': /* --no-dereference */ > symlink_deref_bits = FTS_PHYSICAL; > break; I'm afraid my example was wrong. If I call du -DP I get symlink_deref_bits=FTS_PHYSICAL, but if I call du -PD I get symlink_dref_bits=FTS_COMFOLLOW|FTS_PHYSICAL. -L and -P are not independent (as intended); -D (eventually known as -H) *should* be independent. (-L internally implies -D, but applications don't have to and shouldn't deal with this). -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
