3 URGENT ---------------------------------------------------------------
6 IMPORTANT ------------------------------------------------------------
10 Part of the regression suite should be making sure that we don't
11 break backwards compatibility: old clients vs new servers and so
12 on. Ideally we would test the cross product of versions.
14 It might be sufficient to test downloads from well-known public
15 rsync servers running different versions of rsync. This will give
16 some testing and also be the most common case for having different
17 versions and not being able to upgrade.
21 If the platform doesn't support it, then don't even try.
23 If running as non-root, then don't fail, just give a warning.
24 (There was a thread about this a while ago?)
26 http://lists.samba.org/pipermail/rsync/2001-August/thread.html
27 http://lists.samba.org/pipermail/rsync/2001-September/thread.html
31 Avoids traversal. Better option than a pile of --include statements
32 for people who want to generate the file list using a find(1)
37 Traverse just one directory at a time. Tridge says it's possible.
39 At the moment rsync reads the whole file list into memory at the
40 start, which makes us use a lot of memory and also not pipeline
41 network access as much as we could.
43 We need to be careful of duplicate names getting into the file list.
44 See clean_flist. This could happen if multiple arguments include
47 I think duplicates are only a problem if they're both flowing
48 through the pipeline at the same time. For example we might have
49 updated the first occurrence after reading the checksums for the
50 second. So possibly we just need to make sure that we don't have
51 both in the pipeline at the same time.
53 Possibly if we did one directory at a time that would be sufficient.
55 Alternatively we could pre-process the arguments to make sure no
56 duplicates will ever be inserted.
58 We could have a hash table.
62 At exit, show how much memory was used for the file list, etc.
64 Also we do a wierd exponential-growth allocation in flist.c. I'm
65 not sure this makes sense with modern mallocs. At any rate it will
66 make us allocate a huge amount of memory for large file lists.
70 At the moment hardlink handling is very expensive, so it's off by
71 default. It does not need to be so.
73 We can rule out hardlinked directories since they will probably
74 screw us up in all kinds of ways. They simply should not be used.
76 At the moment rsync only cares about hardlinks to regular files. I
77 guess you could also use them for sockets, devices and other beasts,
78 but I have not seen them.
80 When trying to reproduce hard links, we only need to worry about
81 files that have more than one name (nlinks>1 && !S_ISDIR).
83 The basic point of this is to discover alternate names that refer to
84 the same file. All operations, including creating the file and
85 writing modifications to it need only to be done for the first name.
86 For all later names, we just create the link and then leave it
89 If hard links are to be preserved:
91 Before the generator/receiver fork, the list of files is received
92 from the sender (recv_file_list), and a table for detecting hard
95 The generator looks for hard links within the file list and does
96 not send checksums for them, though it does send other metadata.
98 The sender sends the device number and inode with file entries, so
99 that files are uniquely identified.
101 The receiver goes through and creates hard links (do_hard_links)
102 after all data has been written, but before directory permissions
105 At the moment device and inum are sent as 4-byte integers, which
106 will probably cause problems on large filesystems. On Linux the
107 kernel uses 64-bit ino_t's internally, and people will soon have
108 filesystems big enough to use them. We ought to follow NFS4 in
109 using 64-bit device and inode identification, perhaps with a
110 protocol version bump.
112 Once we've seen all the names for a particular file, we no longer
113 need to think about it and we can deallocate the memory.
115 We can also have the case where there are links to a file that are
116 not in the tree being transferred. There's nothing we can do about
117 that. Because we rename the destination into place after writing,
118 any hardlinks to the old file are always going to be orphaned. In
119 fact that is almost necessary because otherwise we'd get really
120 confused if we were generating checksums for one name of a file and
123 At the moment the code seems to make a whole second copy of the file
124 list, which seems unnecessary.
126 We should have a test case that exercises hard links. Since it
127 might be hard to compare ./tls output where the inodes change we
128 might need a little program to check whether several names refer to
133 Implement suggestions from http://www.kame.net/newsletter/19980604/
134 and ftp://ftp.iij.ad.jp/pub/RFC/rfc2553.txt
136 If a host has multiple addresses, then listen try to connect to all
137 in order until we get through. (getaddrinfo may return multiple
138 addresses.) This is kind of implemented already.
140 Possibly also when starting as a server we may need to listen on
141 multiple passive addresses. This might be a bit harder, because we
142 may need to select on all of them. Hm.
144 Define a syntax for IPv6 literal addresses. Since they include
145 colons, they tend to break most naming systems, including ours.
146 Based on the HTTP IPv6 syntax, I think we should use
148 rsync://[::1]/foo/bar
151 which should just take a small change to the parser code.
155 If we hang or get SIGINT, then explain where we were up to. Perhaps
156 have a static buffer that contains the current function name, or
157 some kind of description of what we were trying to do. This is a
158 little easier on people than needing to run strace/truss.
160 "The dungeon collapses! You are killed." Rather than "unexpected
161 eof" give a message that is more detailed if possible and also more
166 Device major/minor numbers should be at least 32 bits each. See
167 http://lists.samba.org/pipermail/rsync/2001-November/005357.html
169 Transfer ACLs. Need to think of a standard representation.
170 Probably better not to even try to convert between NT and POSIX.
171 Possibly can share some code with Samba.
175 With the current common --include '*/' --exclude '*' pattern, people
176 can end up with many empty directories. We might avoid this by
177 lazily creating such directories.
181 Perhaps don't use our own zlib. Will we actually be incompatible,
182 or just be slightly less efficient?
186 Perhaps flush stdout after each filename, so that people trying to
187 monitor progress in a log file can do so more easily. See
188 http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=48108
192 There are already some patches to do this.
194 PLATFORMS ------------------------------------------------------------
198 Don't detach, because this messes up --srvany.
200 http://sources.redhat.com/ml/cygwin/2001-08/msg00234.html
202 According to "Effective TCP/IP Programming" (??) close() on a socket
203 has incorrect behaviour on Windows -- it sends a RST packet to the
204 other side, which gives a "connection reset by peer" error. On that
205 platform we should probably do shutdown() instead. However, on Unix
206 we are correct to call close(), because shutdown() discards
209 DOCUMENTATION --------------------------------------------------------
213 BUILD FARM -----------------------------------------------------------
217 AMDAHL UTS (Dave Dykstra)
219 Cygwin (on different versions of Win32?)
221 HP-UX variants (via HP?)
225 NICE -----------------------------------------------------------------
229 Re-read config file (just exec() ourselves) rather than exiting.
231 --no-detach and --no-fork options
233 Very useful for debugging. Also good when running under a
234 daemon-monitoring process that tries to restart the service when the
237 hang/timeout friendliness
243 Indicate whether files are new, updated, or deleted
247 Change to using gettext(). Probably need to ship this for platforms
250 Solicit translations.
256 Write a small emulation of interactive ftp as a Pythonn program
257 that calls rsync. Commands such as "cd", "ls", "ls *.c" etc map
258 fairly directly into rsync commands: it just needs to remember the
259 current host, directory and so on. We can probably even do
260 completion of remote filenames.