Now that the long-standing chgrp bug is fixed, go back to -vvv from -vvvv.
[rsync/rsync.git] / NEWS
... / ...
CommitLineData
1NEWS for rsync 2.6.1 (UNRELEASED)
2Protocol: 27 (unchanged)
3Changes since 2.6.0:
4
5 ENHANCEMENTS:
6
7 * Lower memory use and more optimal transfer of data over
8 the socket (see the INTERNAL section for details).
9
10 * The RSYNC_PROXY environment variable can now contain a
11 "USER:PASS@" prefix before the "HOST:PORT" information.
12 (Bardur Arantsson)
13
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.
17
18 BUG FIXES:
19
20 * The --link-dest code now works properly for a non-root user
21 when the UIDs of the source and destination differ and -o was
22 specified, and when the group of the source can't be used on
23 the destination and -g was specified. (Wayne Davison)
24
25 * Fixed a bug in the handling of -H (hard-links) that might
26 cause the expanded PATH/NAME value of the current item to
27 get overwritten (due to an expanded-name caching bug).
28 (Wayne Davison)
29
30 * We now reset the "new data has been sent" flag at the start
31 of each file we send. This makes sure that an interrupted
32 transfer with the --partial option set doesn't keep a shorter
33 temp file than the current basis file when no new data has been
34 transfered over the wire for that file. (Wayne Davison)
35
36 * Fixed a byte-order problem in --batch-mode on big-endian
37 machines. (Jay Fenlason)
38
39 * Fixed configure bug when running "./configure --disable-ipv6".
40 (Wayne Davison)
41
42 * Fixed "make test" bug when build dir is not the source dir.
43
44 * When using --cvs-exclude, the exclude items we get from a
45 directory's .cvsignore file once again only affect that one
46 directory (and not all following directories too).
47
48 * When transferring a file that has group 0 with -g specified
49 (typically via -a) and not enough privs to retain the group,
50 rsync no longer complains about chown failing.
51
52 * When specifying the USER@HOST: prefix for a file, the USER
53 part can now contain an '@', if needed (i.e. the last '@'
54 is used to find the HOST, not the first).
55
56 INTERNAL:
57
58 * Most of the I/O is now buffered, which results in a pretty
59 large speedup when running under MS Windows. (Craig Barratt)
60
61 * Optimizations to the name-handling/comparing code have made
62 some significant reductions in user-CPU time for large file
63 sets. (Wayne Davison)
64
65 * Some variable-type cleanup that makes the code more consistent.
66 (Wayne Davison)
67
68 * Reduced memory requirements of hard link preservation.
69 (J.W. Schultz)
70
71 * Implemented a new algorithm for hard-link handling that speeds
72 up the code significantly. (J.W. Schultz and Wayne Davison)
73
74 * The --hard-link option now uses the first existing file in the
75 group of linked files as the basis for the transfer. This
76 prevents the sub-optimal transfer of a file's data when a new
77 hardlink is added on the sending side and it sorts alphabetically
78 earlier in the list than the files that are already present on the
79 receiving side.
80
81 * Got rid of support for protocol versions 17 and 18 (which are
82 both over 6 years old). (Wayne Davison)
83
84 * More optimal data transmission for --hard-links (protocol 28).
85
86 * More optimal data transmission for --checksum (protocol 28).
87
88 * Less memory is used when --checksum is specified.
89
90 * Less memory is used in the file list (a per-file savings).
91
92 * The 2 pipes used between the receiver and generator processes
93 (which are forked on the same machine) were reduced to 1 pipe
94 and the protocol improved so that (1) it is now impossible to
95 have the "redo" pipe fill up and hang rsync, and (2) trailing
96 messages from the receiver don't get lost on their way through
97 the generator over to the sender (the latter mainly affected
98 hard-link messages). (Wayne Davison)
99
100 * The reading & writing of the file list in batch-mode is now
101 handled by the same code that sends & receives the list over
102 the wire. This makes it much easier to maintain.
103
104 * Optimized the -x (--one-file-system) flag's handling of any
105 mount-point directories we encounter (it no longer scans the
106 contents of the mount-point dirs, just to throw away the data).
107
108\f
109NEWS for rsync 2.6.0 (1 Jan 2004)
110Protocol: 27 (changed)
111Changes since 2.5.7:
112
113 ENHANCEMENTS:
114
115 * "ssh" is now the default remote shell for rsync. If you want to
116 change this, configure like this: "./configure --with-rsh=rsh".
117
118 * Added --files-from, --no-relative, --no-implied-dirs, and --from0.
119 Note that --from0 affects the line-ending character for all the
120 files read by the --*-from options. (Wayne Davison)
121
122 * Length of csum2 is now per-file starting with protocol version
123 27. (J.W. Schultz)
124
125 * Per-file dynamic block size is now sqrt(file length). The
126 per-file checksum size is determined according to an algorithm
127 provided by Donovan Baarda which reduces the probability of rsync
128 algorithm corrupting data and falling back using the whole md4
129 checksums. (J.W. Schultz, Donovan Baarda)
130
131 * The --stats option no longer includes the (debug) malloc summary
132 unless the verbose option was specified at least twice.
133
134 * Added a new error/warning code for when files vanish from the
135 sending side. Made vanished source files not interfere with the
136 file-deletion pass when --delete-after was specified.
137
138 * Various trailing-info sections are now preceded by a newline.
139
140 BUG FIXES:
141
142 * Fixed several exclude/include matching bugs when using wild-cards.
143 This has a several user-visible effects, all of which make the
144 matching more consistent and intuitive. This should hopefully not
145 cause anyone problems since it makes the matching work more like
146 what people are expecting. (Wayne Davison)
147
148 - A pattern with a "**" no longer causes a "*" to match slashes.
149 For example, with "/*/foo/**", "foo" must be 2 levels deep.
150 [If your string has BOTH "*" and "**" wildcards, changing the
151 "*" wildcards to "**" will provide the old behavior in all
152 versions.]
153
154 - "**/foo" now matches at the base of the transfer (like /foo
155 does). [Use "/**/foo" to get the old behavior in all versions.]
156
157 - A non-anchored wildcard term floats to match beyond the base of
158 the transfer. E.g. "CVS/R*" matches at the end of the path,
159 just like the non-wildcard term "CVS/Root" does. [Use "/CVS/R*"
160 to get the old behavior in all versions.]
161
162 - Including a "**" in the match term causes it to be matched
163 against the entire path, not just the name portion, even if
164 there aren't any interior slashes in the term. E.g. "foo**bar"
165 would exclude "/path/foo-bar" (just like before) as well as
166 "/foo-path/baz-bar" (unlike before). [Use "foo*bar" to get the
167 old behavior in all versions.]
168
169 * The exclude list specified in the daemon's config file is now
170 properly applied to the pulled items no matter how deep the
171 user's file-args are in the source tree. (Wayne Davison)
172
173 * For protocol version >= 27, mdfour_tail() is called when the
174 block size (including checksum_seed) is a multiple of 64.
175 Previously it was not called, giving the wrong MD4 checksum.
176 (Craig Barratt)
177
178 * For protocol version >= 27, a 64 bit bit counter is used in
179 mdfour.c as required by the RFC. Previously only a 32 bit bit
180 counter was used, causing incorrect MD4 file checksums for
181 file sizes >= 512MB - 4. (Craig Barratt)
182
183 * Fixed a crash bug when interacting with older rsync versions and
184 multiple files of the same name are destined for the same dir.
185 (Wayne Davison)
186
187 * Keep tmp names from overflowing MAXPATHLEN.
188
189 * Make --link-dest honor the absence of -p, -o, and -g.
190
191 * Made rsync treat a trailing slash in the destination in a more
192 consistent manner.
193
194 * Fixed file I/O error detection. (John Van Essen)
195
196 * Fixed bogus "malformed address {hostname}" message in rsyncd log
197 when checking IP address against hostnames from "hosts allow"
198 and "hosts deny" parameters in config file.
199
200 * Print heap statistics when verbose >= 2 instead of when >= 1.
201
202 * Fixed a compression (-z) bug when syncing a mostly-matching file
203 that contains already-compressed data. (Yasuoka Masahiko and
204 Wayne Davison)
205
206 * Fixed a bug in the --backup code that could cause deleted files
207 to not get backed up.
208
209 * When the backup code makes new directories, create them with mode
210 0700 instead of 0755 (since the directory permissions in the
211 backup tree are not yet copied from the main tree).
212
213 * Call setgroups() in a more portable manner.
214
215 * Improved file-related error messages to better indicate exactly
216 what pathname failed. (Wayne Davison)
217
218 * Fixed some bugs in the handling of --delete and --exclude when
219 using the --relative (-R) option. (Wayne Davison)
220
221 * Fixed bug that prevented regular files from replacing
222 special files and caused a directory in --link-dest or
223 --compare-dest to block the creation of a file with the
224 same path. A directory still cannot be replaced by a
225 regular file unless --delete specified. (J.W. Schultz)
226
227 * Detect and report when open or opendir succeed but read and
228 readdir fail caused by network filesystem issues and truncated
229 files. (David Norwood, Michael Brown, J.W. Schultz)
230
231 * Added a fix that should give ssh time to restore the tty settings
232 if the user presses Ctrl-C at an ssh password prompt.
233
234 INTERNAL:
235
236 * Eliminated vestigial support for old versions that we stopped
237 supporting. (J.W. Schultz)
238
239 * Simplified some of the option-parsing code. (Wayne Davison)
240
241 * Some cleanup made to the exclude code, as well as some new
242 defines added to enhance readability. (Wayne Davison)
243
244 * Changed the protocol-version code so that it can interact at a
245 lower protocol level than the maximum supported by both sides.
246 Added an undocumented option, --protocol=N, to force the value
247 we advertise to the other side (primarily for testing purposes).
248 (Wayne Davison)