1 NEWS for rsync 2.6.1 (UNRELEASED)
2 Protocol: 27 (unchanged)
7 * Lower memory use and more optimal transfer of data over
8 the socket (see the INTERNAL section for details).
10 * The RSYNC_PROXY environment variable can now contain a
11 "USER:PASS@" prefix before the "HOST:PORT" information.
14 * The --progress output now mentions how far along in the
15 transfer we are, including both a count of files transferred
16 and a percentage of the total file-count that we're processed.
20 * When -x (--one-file-system) is combined with -L (--copy-links)
21 or --copy-unsafe-links, no symlinked files are skipped, even
22 if the referant file is on a different filesystem.
24 * The --link-dest code now works properly for a non-root user
25 when the UIDs of the source and destination differ and -o was
26 specified, and when the group of the source can't be used on
27 the destination and -g was specified. (Wayne Davison)
29 * Fixed a bug in the handling of -H (hard-links) that might
30 cause the expanded PATH/NAME value of the current item to
31 get overwritten (due to an expanded-name caching bug).
34 * We now reset the "new data has been sent" flag at the start
35 of each file we send. This makes sure that an interrupted
36 transfer with the --partial option set doesn't keep a shorter
37 temp file than the current basis file when no new data has been
38 transfered over the wire for that file. (Wayne Davison)
40 * Fixed a byte-order problem in --batch-mode on big-endian
41 machines. (Jay Fenlason)
43 * Fixed configure bug when running "./configure --disable-ipv6".
46 * Fixed "make test" bug when build dir is not the source dir.
48 * When using --cvs-exclude, the exclude items we get from a
49 directory's .cvsignore file once again only affect that one
50 directory (and not all following directories too).
52 * When transferring a file that has group 0 with -g specified
53 (typically via -a) and not enough privs to retain the group,
54 rsync no longer complains about "chown" failing.
56 * When specifying the USER@HOST: prefix for a file, the USER
57 part can now contain an '@', if needed (i.e. the last '@'
58 is used to find the HOST, not the first).
60 * Fixed some bugs in the handling of group IDs for non-root
61 users: (1) It properly handles a group that the sender didn't
62 have a name for (it would previously skip changing the group
63 on any files in that group). (2) If --numeric-ids is used,
64 rsync no longer attempts to set groups that the user doesn't
65 have the permission to set.
69 * Most of the I/O is now buffered, which results in a pretty
70 large speedup when running under MS Windows. (Craig Barratt)
72 * Optimizations to the name-handling/comparing code have made
73 some significant reductions in user-CPU time for large file
76 * Some variable-type cleanup that makes the code more consistent.
79 * Reduced memory requirements of hard link preservation.
82 * Implemented a new algorithm for hard-link handling that speeds
83 up the code significantly. (J.W. Schultz and Wayne Davison)
85 * The --hard-link option now uses the first existing file in the
86 group of linked files as the basis for the transfer. This
87 prevents the sub-optimal transfer of a file's data when a new
88 hardlink is added on the sending side and it sorts alphabetically
89 earlier in the list than the files that are already present on the
92 * Dropped support for protocol versions less than 20
93 (2.3.0 released 15 Mar 1999) and activated warnings for
94 protocols less than 25 (2.5.0 released 23 Aug 2001)
95 (Wayne Davison and J.W. Schultz, severally)
97 * More optimal data transmission for --hard-links (protocol 28).
99 * More optimal data transmission for --checksum (protocol 28).
101 * Less memory is used when --checksum is specified.
103 * Less memory is used in the file list (a per-file savings).
105 * Changed hardlink info and file_struct + strings to use
106 allocation pools. This reduces memory use for large
107 filesets and permits freeing memory to the OS. (J.W. Schultz)
109 * The 2 pipes used between the receiver and generator processes
110 (which are forked on the same machine) were reduced to 1 pipe
111 and the protocol improved so that (1) it is now impossible to
112 have the "redo" pipe fill up and hang rsync, and (2) trailing
113 messages from the receiver don't get lost on their way through
114 the generator over to the sender (the latter mainly affected
115 hard-link messages). (Wayne Davison)
117 * The reading & writing of the file list in batch-mode is now
118 handled by the same code that sends & receives the list over
119 the wire. This makes it much easier to maintain.
121 * Optimized the -x (--one-file-system) flag's handling of any
122 mount-point directories we encounter (it no longer scans the
123 contents of the mount-point dirs, just to throw away the data).
125 * Improved the internal uid/gid code to be more portable and
126 a little more optimized.
129 NEWS for rsync 2.6.0 (1 Jan 2004)
130 Protocol: 27 (changed)
135 * "ssh" is now the default remote shell for rsync. If you want to
136 change this, configure like this: "./configure --with-rsh=rsh".
138 * Added --files-from, --no-relative, --no-implied-dirs, and --from0.
139 Note that --from0 affects the line-ending character for all the
140 files read by the --*-from options. (Wayne Davison)
142 * Length of csum2 is now per-file starting with protocol version
145 * Per-file dynamic block size is now sqrt(file length). The
146 per-file checksum size is determined according to an algorithm
147 provided by Donovan Baarda which reduces the probability of rsync
148 algorithm corrupting data and falling back using the whole md4
149 checksums. (J.W. Schultz, Donovan Baarda)
151 * The --stats option no longer includes the (debug) malloc summary
152 unless the verbose option was specified at least twice.
154 * Added a new error/warning code for when files vanish from the
155 sending side. Made vanished source files not interfere with the
156 file-deletion pass when --delete-after was specified.
158 * Various trailing-info sections are now preceded by a newline.
162 * Fixed several exclude/include matching bugs when using wild-cards.
163 This has a several user-visible effects, all of which make the
164 matching more consistent and intuitive. This should hopefully not
165 cause anyone problems since it makes the matching work more like
166 what people are expecting. (Wayne Davison)
168 - A pattern with a "**" no longer causes a "*" to match slashes.
169 For example, with "/*/foo/**", "foo" must be 2 levels deep.
170 [If your string has BOTH "*" and "**" wildcards, changing the
171 "*" wildcards to "**" will provide the old behavior in all
174 - "**/foo" now matches at the base of the transfer (like /foo
175 does). [Use "/**/foo" to get the old behavior in all versions.]
177 - A non-anchored wildcard term floats to match beyond the base of
178 the transfer. E.g. "CVS/R*" matches at the end of the path,
179 just like the non-wildcard term "CVS/Root" does. [Use "/CVS/R*"
180 to get the old behavior in all versions.]
182 - Including a "**" in the match term causes it to be matched
183 against the entire path, not just the name portion, even if
184 there aren't any interior slashes in the term. E.g. "foo**bar"
185 would exclude "/path/foo-bar" (just like before) as well as
186 "/foo-path/baz-bar" (unlike before). [Use "foo*bar" to get the
187 old behavior in all versions.]
189 * The exclude list specified in the daemon's config file is now
190 properly applied to the pulled items no matter how deep the
191 user's file-args are in the source tree. (Wayne Davison)
193 * For protocol version >= 27, mdfour_tail() is called when the
194 block size (including checksum_seed) is a multiple of 64.
195 Previously it was not called, giving the wrong MD4 checksum.
198 * For protocol version >= 27, a 64 bit bit counter is used in
199 mdfour.c as required by the RFC. Previously only a 32 bit bit
200 counter was used, causing incorrect MD4 file checksums for
201 file sizes >= 512MB - 4. (Craig Barratt)
203 * Fixed a crash bug when interacting with older rsync versions and
204 multiple files of the same name are destined for the same dir.
207 * Keep tmp names from overflowing MAXPATHLEN.
209 * Make --link-dest honor the absence of -p, -o, and -g.
211 * Made rsync treat a trailing slash in the destination in a more
214 * Fixed file I/O error detection. (John Van Essen)
216 * Fixed bogus "malformed address {hostname}" message in rsyncd log
217 when checking IP address against hostnames from "hosts allow"
218 and "hosts deny" parameters in config file.
220 * Print heap statistics when verbose >= 2 instead of when >= 1.
222 * Fixed a compression (-z) bug when syncing a mostly-matching file
223 that contains already-compressed data. (Yasuoka Masahiko and
226 * Fixed a bug in the --backup code that could cause deleted files
227 to not get backed up.
229 * When the backup code makes new directories, create them with mode
230 0700 instead of 0755 (since the directory permissions in the
231 backup tree are not yet copied from the main tree).
233 * Call setgroups() in a more portable manner.
235 * Improved file-related error messages to better indicate exactly
236 what pathname failed. (Wayne Davison)
238 * Fixed some bugs in the handling of --delete and --exclude when
239 using the --relative (-R) option. (Wayne Davison)
241 * Fixed bug that prevented regular files from replacing
242 special files and caused a directory in --link-dest or
243 --compare-dest to block the creation of a file with the
244 same path. A directory still cannot be replaced by a
245 regular file unless --delete specified. (J.W. Schultz)
247 * Detect and report when open or opendir succeed but read and
248 readdir fail caused by network filesystem issues and truncated
249 files. (David Norwood, Michael Brown, J.W. Schultz)
251 * Added a fix that should give ssh time to restore the tty settings
252 if the user presses Ctrl-C at an ssh password prompt.
256 * Eliminated vestigial support for old versions that we stopped
257 supporting. (J.W. Schultz)
259 * Simplified some of the option-parsing code. (Wayne Davison)
261 * Some cleanup made to the exclude code, as well as some new
262 defines added to enhance readability. (Wayne Davison)
264 * Changed the protocol-version code so that it can interact at a
265 lower protocol level than the maximum supported by both sides.
266 Added an undocumented option, --protocol=N, to force the value
267 we advertise to the other side (primarily for testing purposes).