LineageOS/android_external_rsync
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity
5,312 Commits 2 Branches 0 Tags
f7a76b9c450c05abb799e1103d90b51f9f06c0c3
Commit Graph

378 Commits

Author SHA1 Message Date
Wayne Davison
184dd27a18 Call delete_in_dir(NULL, NULL, NULL) in do_delete_pass(). 2005-11-09 19:46:42 +00:00
Wayne Davison
0e5665d3ab Made recursively deleting extraneous directories more memory 
efficient by removing the need for a MAXPATHLEN buffer on the
stack of delete_item().
 2005-11-09 19:39:46 +00:00
Wayne Davison
0fdb1aa8ff In the device section, avoid calling unlink() if the stat() failed. 2005-10-30 22:28:11 +00:00
Wayne Davison
02b5cb238b Added code to implement the --min-size logic. 2005-10-14 20:16:09 +00:00
Wayne Davison
e90aab4982 Changed the variable names for --existing and --ignore-existing. 2005-09-29 18:07:55 +00:00
Wayne Davison
991daf008d Keeping silent about stat() errors is never a good idea -- it hides 
files that are not up-to-date.
 2005-09-29 17:07:45 +00:00
Wayne Davison
a5b786d80f Tweaked the calculation that goes into sum->count so that it cannot 
overflow into a negative value (which is particularly important if
configure didn't find a real int64 type).
 2005-09-06 18:12:38 +00:00
Wayne Davison
15cf186b85 Don't recreate a device if just the permissions bits have changed 
in the st_mode.
 2005-08-01 04:09:58 +00:00
Wayne Davison
273a7ed59f Fixed the handling of hard-linked symlinks and devices. 2005-07-31 23:18:28 +00:00
Wayne Davison
6cc1198288 Support new --append option. 2005-07-28 01:46:12 +00:00
Wayne Davison
00b9618460 - Fixed the mode comparison in itemize() to mask off the bits 
outside of the CHMOD_BITS mask.
- Don't call do_chmod() unless HAVE_CHMOD is defined.
- Don't mask the mode bits being sent to do_chmod().
 2005-07-27 23:30:55 +00:00
Wayne Davison
092906933b Ignore EEXIST on second mkdir() call. 2005-06-30 17:03:14 +00:00
Wayne Davison
6ff2f4ba67 Use cmp_modtime() in a couple more spots where a local modtime is 
compared with a remote modtime.
 2005-06-28 00:12:08 +00:00
Wayne Davison
215b444cae Don't set ITEM_REPORT_TIME when a hard-linked file is already 
linked up as it should be and --times was not enabled.
 2005-06-09 21:57:45 +00:00
Wayne Davison
26beb7861f - Don't stop deleting in a dir if one item fails. 
- Don't skip deleting in a dir if the initial rmdir fails
  (2.6.4 only continued if it was a non-empty error).
 2005-05-22 20:52:32 +00:00
Wayne Davison
6e8b9f1341 Must check FLAG_MOUNT_POINT in delete_in_dir() to avoid starting a 
descent into a mounted directory when -x is specified.
 2005-05-22 17:19:16 +00:00
Wayne Davison
ac40b74788 The global allowed_lull is now already set, so just set lull_mod. 2005-05-19 08:52:22 +00:00
Wayne Davison
4d474ad513 One place that was testing errno for ENOTEMPTY was failing to 
also test for EEXIST (which is returned by some OSes).
 2005-05-19 00:00:28 +00:00
Wayne Davison
b225b089b8 Have do_delete_pass() immediately return if list_only is set. 2005-05-13 23:00:20 +00:00
Wayne Davison
289a32167c Moved the max-size checking code down so that it only checks 
regular files.
 2005-04-28 16:26:23 +00:00
Wayne Davison
513fd04d21 Set ignore_timeout before starting the transfer, regardless of 
what protocol level is in effect.  This guards against the
generator timing out when the output pipe is full and the input
pipe (to the receiver) is slow.
 2005-04-27 22:58:06 +00:00
Wayne Davison
468d766819 Simplified the last change a little using a fix derived from 
Mike Castle's suggestion.
 2005-04-24 22:34:20 +00:00
Wayne Davison
f3ab64d3a8 When deleting files, we need to take any local FLAG_MOUNT_POINT 
flag-bit and put it into the regular flist's flags for that dir,
and delete_in_dir() now refuses to delete a directory that has
that bit set.
 2005-04-24 22:11:26 +00:00
Wayne Davison
3556fe5d86 Fixed a problem where a --link-dest hard-link could cause us to 
miss out on the hard_link_cluster() call.
 2005-04-23 19:46:22 +00:00
Wayne Davison
d8169e6f6b - Call hard_link_check() with its new args. 
- Got rid of a duplicate error after the hard_link_one() call.
 2005-04-23 17:48:31 +00:00
Wayne Davison
2a5d5a8cc4 If we're outputting a message about the remote file in a 
single-file transfer, we need to make sure that the file's
name is the generic-transfer name and not a local name.
 2005-04-22 16:45:33 +00:00
Wayne Davison
15164c0aa9 Paul's patch to improve the "not creating new ..." message 
for directories.
 2005-04-10 19:44:16 +00:00
Wayne Davison
beb51aa09e Use the new do_xfers variable in place of some dry_run checks 
(with appropriate negation).
 2005-04-09 18:59:52 +00:00
Wayne Davison
c2523a0541 Don't call maybe_flush_socket() quite so often. 2005-04-07 08:32:32 +00:00
Wayne Davison
cd908ef4ff Fix a bug with --delete-after combined with --backup. 2005-04-07 08:13:44 +00:00
Wayne Davison
417099fa20 Periodically call maybe_flush_socket(). 2005-04-05 20:08:51 +00:00
Wayne Davison
a06e2b7cab When doing a delete pass with do_progress output, get rid of the 
temporary output (it used to always be covered up by output in
older versions, but it might be followed by a newline in newer
versions).
 2005-04-05 19:51:13 +00:00
Wayne Davison
552a218468 Make a local-copy caused by a not-quite-up-to-date --compare-dest 
or --link-dest file be output as a 'c' (local change) when itemizing
or as transfered file when not itemizing.
 2005-04-01 18:12:22 +00:00
Wayne Davison
41b84ce012 Don't try to tweak read-only dest-dir permissions in list-only mode. 2005-03-31 23:30:03 +00:00
Wayne Davison
9c63d83d3d Got rid of a useless extern. 2005-03-30 16:33:41 +00:00
Wayne Davison
083acd496d Turn off do_progress during the file-update phase so that 
--delete-during and --fuzzy don't get any superfluous
" N files..." lines output in the middle of the processing.
 2005-03-29 10:41:26 +00:00
Wayne Davison
70352269ba If --delay-updates wasn't specified, write out two -1 indexes at 
the same time to avoid a useless round-trip delay for an empty
delay-updates phase.
 2005-03-29 01:04:20 +00:00
Wayne Davison
94af17a6eb Got rid of temporary code that set the ITEM_DUMMY_BIT. 2005-03-27 05:58:36 +00:00
Wayne Davison
f328e0f3a8 Set a maximum distance-measure that find_fuzzy() will accept. 2005-03-17 08:59:48 +00:00
Wayne Davison
0438f100ae We need to run our post-processing activities after the end of 
the receiver's delay-update processing.
 2005-03-17 00:41:18 +00:00
Wayne Davison
62f9573fb3 - Added an extra phase to the end of the transfer to better handle 
delayed updates that have hard links.
- Send the new ITEM_DUMMY_BIT to the sender so that we can figure
  out if the other side is pre1 or pre2 and let the receiver reject
  it.
 2005-03-15 19:19:38 +00:00
Wayne Davison
7e9059d60f Fixed a just-introduced crash bug in the --fuzzy processing. 2005-03-14 22:22:42 +00:00
Wayne Davison
ccc51c8331 The --fuzzy code now handles a file->dirname that has an identical 
string as another file without being an identical pointer.
 2005-03-14 03:35:40 +00:00
Wayne Davison
717b04306a Tweaked the name of a variable. 2005-03-12 23:52:08 +00:00
Wayne Davison
967866d4df Added --copy-dest logic, and improved the updating of --compare-dest 
and --link-dest files that are up-to-date but have differing attributes.
 2005-03-11 17:36:03 +00:00
Wayne Davison
1f1d368ad5 - Improved the error-checking for some delete_item() calls. 
- Move the non-regular-file delete-check above the alt-basis check
  where it belongs.
- Keep track of the real statret and real stat-struct for certain
  alt-basis scenarios (e.g. partial-dir and fuzzy) so that we send
  the right itemized change flags.
 2005-03-09 23:46:28 +00:00
Wayne Davison
ef20efcbb6 Made itemize() output the fnamecmp_type and the fuzzy name based on 
the new ITEM_BASIS_TYPE_FOLLOWS and ITEM_XNAME_FOLLOWS flags.  Got
rid of the name-pipe to the receiver.
 2005-03-09 18:54:12 +00:00
Wayne Davison
1f7e29b99c Fixed the change-report output for a directory that has no 
write permissions.
 2005-03-09 03:49:07 +00:00
Wayne Davison
c2b11ba017 Backed out the hack that reversed ITEM_REPORT_XATTRS with 
ITEM_TRANSFER.  Yes, it allowed some kludge code that made backward
compatibility seamless, but it made it impossible to remove the hack
in the future.  This way, the backward compatibility is just
slightly inaccurate in the display of the first letter in the %i
output, and the only hack can be safely removed without causing
problems.
 2005-03-06 23:37:42 +00:00
Wayne Davison
f75a53e71b - When --max-delete is exceeded, we now count how many deletions 
would have happend, warn about the number skipped, and set
  io_error to IOERR_DEL_LIMIT.
- When dry_run > 1 (which indicates that the destinationdir is
  missing), skip deletions in that dir.  This fixes a bug in a
  copy that is creating the destination dir w/--delete enabled.
 2005-03-05 18:58:42 +00:00