If --backup was specified without a --backup-dir, don't preserve the
[rsync/rsync.git] / NEWS
... / ...
CommitLineData
1NEWS for rsync 2.6.3 (UNRELEASED)
2Protocol: 28 (unchanged)
3Changes since 2.6.2:
4
5 BUG FIXES:
6
7 - Fixed a crash bug that might appear when --delete was used and
8 multiple source directories were specified.
9
10 - The --backup code no longer attempts to create some directories
11 over and over again (generating warnings along the way).
12
13 - Fixed a bug in the reading of the secrets file (by the daemon) and
14 the password file (by the client): the files no longer need to be
15 terminated by a newline for their content to be read in.
16
17 - If a file has a read error on the sending side, the receiver will
18 no longer keep the resulting file unless the --partial option was
19 specified.
20
21 - Fixed an age-old crash problem with --read-batch on a local copy
22 (rsync was improperly assuming --whole-file for the local copy).
23
24 - Files specified in the daemon's "exclude" or "exclude from" config
25 items are now excluded from being uploaded (if the module is not
26 read-only) as well as downloaded.
27
28 ENHANCEMENTS:
29
30 - Added the "write only" option to the daemon's config file.
31
32 - Added long-option names for -4 and -6 (namely --ipv4 and --ipv6)
33 and documented all these options in the man page.
34
35 INTERNAL:
36
37 - Some cleanup in the exclude code has saved some memory and made
38 things easier to maintain.
39
40 - Use rsyserr() in the various places that were still calling
41 rprintf() with strerror() as an arg.
42
43 BUILD CHANGES:
44
45 - Added a "gen" target to rebuild most of the various generated
46 files, including configure, config.h.in, the man pages, and
47 proto.h.
48
49 - The timestamp of the proto.h file is unaltered by "make proto"
50 if the contents don't change.
51
52 DEVELOPER RELATED:
53
54 - The scripts in the testsuite dir were cleaned up a bit.
55
56NEWS for rsync 2.6.2 (30 Apr 2004)
57Protocol: 28 (unchanged)
58Changes since 2.6.1:
59
60 BUG FIXES:
61
62 - Fixed a major bug in the sorting of the filenames when --relative
63 is used for some sources (just sources such as "/" and "/*" were
64 affected). This fix ensures that we ask for the right file-list
65 item when requesting changes from the sender.
66
67 - Rsync now checks the return value of the close() function to
68 better report disk-full problems on an NFS file system.
69
70 - Restored the old daemon-server behavior of logging error messages
71 rather than returning them to the user. (A better long-term fix
72 will be sought in the future.)
73
74 - An obscure uninitialized-variable bug was fixed in the uid/gid
75 code. (This bug probably had no ill effects.)
76
77 BUILD CHANGES:
78
79 - Got rid of the configure check for sys/sysctl.h (it wasn't used
80 and was causing a problem on some systems). Also improved the
81 broken-largefile-locking test to try to avoid failure due to an
82 NFS build-dir.
83
84 - Fixed a compile problem on systems that don't define
85 AI_NUMERICHOST.
86
87 - Fixed a compile problem in the popt source for compilers that
88 don't support __attribute__.
89
90 DEVELOPER RELATED:
91
92 - Improved the testsuite's "merge" test to work on OSF1.
93
94 - Two new diffs were added to the patches dir.
95
96\f
97NEWS for rsync 2.6.1 (26 Apr 2004)
98Protocol: 28 (changed)
99Changes since 2.6.0:
100
101 SECURITY FIXES:
102
103 - Paths sent to an rsync daemon are more thoroughly sanitized when
104 chroot is not used. If you're running a non-read-only rsync
105 daemon with chroot disabled, *please upgrade*, ESPECIALLY if the
106 user privs you run rsync under is anything above "nobody".
107
108 ENHANCEMENTS:
109
110 - Lower memory use, more optimal transfer of data over the socket,
111 and lower CPU usage (see the INTERNAL section for details).
112
113 - The RSYNC_PROXY environment variable can now contain a
114 "USER:PASS@" prefix before the "HOST:PORT" information.
115 (Bardur Arantsson)
116
117 - The --progress output now mentions how far along in the transfer
118 we are, including both a count of files transferred and a
119 percentage of the total file-count that we've processed. It also
120 shows better current-rate-of-transfer and remaining-transfer-time
121 values.
122
123 - Documentation changes now attempt to describe some often mis-
124 understood features more clearly.
125
126 BUG FIXES:
127
128 - When -x (--one-file-system) is combined with -L (--copy-links) or
129 --copy-unsafe-links, no symlinked files are skipped, even if the
130 referent file is on a different filesystem.
131
132 - The --link-dest code now works properly for a non-root user when
133 (1) the UIDs of the source and destination differ and -o was
134 specified, or (2) when the group of the source can't be used on
135 the destination and -g was specified.
136
137 - Fixed a bug in the handling of -H (hard-links) that might cause
138 the expanded PATH/NAME value of the current item to get
139 overwritten (due to an expanded-name caching bug).
140
141 - We now reset the "new data has been sent" flag at the start of
142 each file we send. This makes sure that an interrupted transfer
143 with the --partial option set doesn't keep a shorter temp file
144 than the current basis file when no new data has been transfered
145 over the wire for that file.
146
147 - Fixed a byte-order problem in --batch-mode on big-endian machines.
148 (Jay Fenlason)
149
150 - When using --cvs-exclude, the exclude items we get from a
151 per-directory's .cvsignore file once again only affect that one
152 directory (not all following directories too). The items are also
153 now properly word-split and parsed without any +/- prefix parsing.
154
155 - When specifying the USER@HOST: prefix for a file, the USER part
156 can now contain an '@', if needed (i.e. the last '@' is used to
157 find the HOST, not the first).
158
159 - Fixed some bugs in the handling of group IDs for non-root users:
160 (1) It properly handles a group that the sender didn't have a name
161 for (it would previously skip changing the group on any files in
162 that group). (2) If --numeric-ids is used, rsync no longer
163 attempts to set groups that the user doesn't have the permission
164 to set.
165
166 - Fixed the "refuse options" setting in the rsyncd.conf file.
167
168 - Improved the -x (--one-file-system) flag's handling of any mount-
169 point directories we encounter. It is both more optimal (in that
170 it no longer does a useless scan of the contents of the mount-
171 point dirs) and also fixes a bug where a remapped mount of the
172 original filesystem could get discovered in a subdir we should be
173 ignoring.
174
175 - Rsync no longer discards a double-slash at the start of a filename
176 when trying to open the file. It also no longer constructs names
177 that start with a double slash (unless the user supplied them).
178
179 - Path-specifying options to a daemon should now work the same with
180 or without chroot turned on. Previously, such a option (such as
181 --link-dest) would get its absolute path munged into a relative
182 one if chroot was not on, making that setting fairly useless.
183 Rsync now transforms the path into one that is based on the
184 module's base dir when chroot is not enabled.
185
186 - Fixed a compatibility problem interacting with older rsync
187 versions that might send us an empty --suffix value without
188 telling us that --backup-dir was specified.
189
190 - The "hosts allow" option for a daemon-over-remote-shell process
191 now has improved support for IPv6 addresses and a fix for systems
192 that have a length field in their socket structs.
193
194 - Fixed the ability to request an empty backup --suffix when sending
195 files to an rsync daemon.
196
197 INTERNAL:
198
199 - Most of the I/O is now buffered, which results in a pretty large
200 speedup when running under MS Windows. (Craig Barratt)
201
202 - Optimizations to the name-handling/comparing code have made some
203 significant reductions in user-CPU time for large file sets.
204
205 - Some cleanup of the variable types make the code more consistent.
206
207 - Reduced memory requirements of hard link preservation.
208 (J.W. Schultz)
209
210 - Implemented a new algorithm for hard-link handling that speeds up
211 the code significantly. (J.W. Schultz and Wayne Davison)
212
213 - The --hard-link option now uses the first existing file in the
214 group of linked files as the basis for the transfer. This
215 prevents the sub-optimal transfer of a file's data when a new
216 hardlink is added on the sending side and it sorts alphabetically
217 earlier in the list than the files that are already present on the
218 receiving side.
219
220 - Dropped support for protocol versions less than 20 (2.3.0 released
221 15 Mar 1999) and activated warnings for protocols less than 25
222 (2.5.0 released 23 Aug 2001). (Wayne Davison and J.W. Schultz,
223 severally)
224
225 - More optimal data transmission for --hard-links (protocol 28).
226
227 - More optimal data transmission for --checksum (protocol 28).
228
229 - Less memory is used when --checksum is specified.
230
231 - Less memory is used in the file list (a per-file savings).
232
233 - The generator is now better about not modifying the file list
234 during the transfer in order to avoid a copy-on-write memory
235 bifurcation (on systems where fork() uses shared memory).
236 Previously, rsync's shared memory would slowly become unshared,
237 resulting in real memory usage nearly doubling on the receiving
238 side by the end of the transfer. Now, as long as permissions
239 are being preserved, the shared memory should remain that way
240 for the entire transfer.
241
242 - Changed hardlink info and file_struct + strings to use allocation
243 pools. This reduces memory use for large file-sets and permits
244 freeing memory to the OS. (J.W. Schultz)
245
246 - The 2 pipes used between the receiver and generator processes
247 (which are forked on the same machine) were reduced to 1 pipe and
248 the protocol improved so that (1) it is now impossible to have the
249 "redo" pipe fill up and hang rsync, and (2) trailing messages from
250 the receiver don't get lost on their way through the generator
251 over to the sender (which mainly affected hard-link messages and
252 verbose --stats output).
253
254 - Improved the internal uid/gid code to be more portable and a
255 little more optimized.
256
257 - The device numbers sent when using --devices are now sent as
258 separate major/minor values with 32-bit accuracy (protocol 28).
259 Previously, the copied devices were sent as a single 32-bit
260 number. This will make inter-operation of 64-bit binaries more
261 compatible with their 32-bit brethren (with both ends of the
262 connection are using protocol 28). Note that optimizations in the
263 binary protocol for sending the device numbers often results in
264 fewer bytes being used than before, even though more precision is
265 now available.
266
267 - Some cleanup of the exclude/include structures and its code made
268 things clearer (internally), simpler, and more efficient.
269
270 - The reading & writing of the file-list in batch-mode is now
271 handled by the same code that sends & receives the list over the
272 wire. This makes it much easier to maintain. (Note that the
273 batch code is still considered to be experimental.)
274
275 BUILD CHANGES:
276
277 - The configure script now accepts --with-rsyncd-conf=PATH to
278 override the default value of the /etc/rsyncd.conf file.
279
280 - Fixed configure bug when running "./configure --disable-ipv6".
281
282 - Fixed compilation problem on Tru64 Unix (having to do with
283 sockaddr.sa_len and sockaddr.sin_len).
284
285 DEVELOPER RELATED:
286
287 - Fixed "make test" bug when build dir is not the source dir.
288
289 - Added a couple extra diffs in the "patches" dir, removed the ones
290 that got applied, and rebuilt the rest.
291