Wayne Davison [Sat, 30 Sep 2006 15:20:17 +0000 (15:20 +0000)]
I decided that combining --link-dest with -I doesn't make much sense
(without some kind of a post-transfer identicality check, at least)
so I'm removing the code that tried to make the two play nice.
Wayne Davison [Sat, 23 Sep 2006 23:15:11 +0000 (23:15 +0000)]
Matt tracked down a bug in the daemon option "incoming chmod" where
a newly created directory would get the chmod permissions that were
specified for a file (via the F previx).
Wayne Davison [Sat, 23 Sep 2006 00:07:28 +0000 (00:07 +0000)]
Improvments by Matt for the --progress option, including updating
the examples to look like the actual output in a modern rsync.
(I did a little rewording too...)
Wayne Davison [Mon, 18 Sep 2006 01:19:26 +0000 (01:19 +0000)]
When parsing a literal IPv6 address in square brackets, pass the
whole string (including the brackets) to the remote shell. (The
old code stripped the leading '[' but left the trailing ']'.)
Wayne Davison [Mon, 18 Sep 2006 00:19:04 +0000 (00:19 +0000)]
Mention the --server and --sender option as being internal,
not-for use by a normal user options. Also mention the
support/rrsync script in this context.
Wayne Davison [Sun, 17 Sep 2006 23:28:19 +0000 (23:28 +0000)]
The --inplace option now notices when the basis data is not coming
from the destination file (or an identical copy) and always writes
out the data (instead of seeking when at an identical offset).
Wayne Davison [Sat, 9 Sep 2006 17:59:02 +0000 (17:59 +0000)]
Added the RSYNC_PID environment variable for the pre-/post-xfer
commands so that the pre-xfer command has a unique ID it can
use to cache information for the post-xfer command.
Wayne Davison [Fri, 1 Sep 2006 00:42:39 +0000 (00:42 +0000)]
Matt pointed out that the code was not properly dropping excludes
that were read from a per-dir merge file when --delete-excluded
is enabled. This additional code fixes that.
Wayne Davison [Sat, 26 Aug 2006 16:02:17 +0000 (16:02 +0000)]
When we're cleaning up a path of "/" in relative mode, we must make
the name "/." (as it was of old) so that there is a filename for us
to send. Fixes a problem with --delete --relative when copying from
the root (/) of the filesystem.
Wayne Davison [Wed, 7 Jun 2006 23:05:17 +0000 (23:05 +0000)]
- Pass "module_id" to the two log-related settings that are no longer
global.
- Added a "restart" arg to log_init() that lets us request that logging
be reinitialized if the log-file/syslog-facility params changed.
Wayne Davison [Thu, 1 Jun 2006 08:04:43 +0000 (08:04 +0000)]
- Got rid of the FNAME logcode enum.
- rwrite() no longer supports newlines at the start of the string.
- maybe_log_item() avoids outputting a non-significant update to
the log file if the log-file-format doesn't include %i.
Wayne Davison [Tue, 30 May 2006 18:26:17 +0000 (18:26 +0000)]
Authorize the current host in the rsyncd.conf file in addition
to localhost and 127.0.0.1 (which will hopefully fix the failing
of the daemon tests on FreeBSD).
Wayne Davison [Tue, 30 May 2006 17:47:20 +0000 (17:47 +0000)]
- Added the --log-file-format option to the daemon section.
- Updated and improved the --out-format section.
- Moved and improved the description of the non-daemon --log-file and
--log-file-format options.
Wayne Davison [Tue, 30 May 2006 17:14:17 +0000 (17:14 +0000)]
- Allow the --log-file-format option to override the "log format"
setting when starting a daemon.
- Make a daemon refuse all --log-file* options.
- Don't use refuse logic on daemon-mode options given to a daemon
module -- reject them via their own message.
Wayne Davison [Mon, 29 May 2006 22:56:48 +0000 (22:56 +0000)]
- Use the renamed stdout_format_has_i, stdout_format_has_o_or_i, and
stdout_format variables.
- In rwrite(), make FNAME output only go to the client.
- Added an arg to log_item() that indicates what kind of output is
desired: FNAME (stdout only), FLOG (log file only), or FINFO (both).
Wayne Davison [Mon, 29 May 2006 22:56:45 +0000 (22:56 +0000)]
- Renamed log_format -> stdout_format.
- Renamed log_format_has_i -> stdout_format_has_i.
- Renamed log_format_has_o_or_i -> stdout_format_has_o_or_i.
- Changed the --log-format to --out-format (though we still accept
the old option, and still send it as well).
- Added the --log-file-format option (and got rid of the code that
made --out-format sometimes affect the log-file's format).