Commit | Line | Data |
---|---|---|
46ef7d1d | 1 | -*- indented-text -*- |
a0365806 | 2 | |
46ef7d1d MP |
3 | URGENT --------------------------------------------------------------- |
4 | ||
33d213bb MP |
5 | |
6 | IMPORTANT ------------------------------------------------------------ | |
7 | ||
8 | Cross-test versions | |
9 | ||
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. | |
13 | ||
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. | |
18 | ||
a2d2e5c0 MP |
19 | use chroot |
20 | ||
21 | If the platform doesn't support it, then don't even try. | |
22 | ||
23 | If running as non-root, then don't fail, just give a warning. | |
24 | (There was a thread about this a while ago?) | |
25 | ||
26 | http://lists.samba.org/pipermail/rsync/2001-August/thread.html | |
27 | http://lists.samba.org/pipermail/rsync/2001-September/thread.html | |
28 | ||
29 | --files-from | |
30 | ||
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) | |
33 | command or a script. | |
34 | ||
35 | Performance | |
36 | ||
37 | Traverse just one directory at a time. Tridge says it's possible. | |
a6a3c3df MP |
38 | |
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. | |
42 | ||
b3e6c815 MP |
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 | |
45 | the same file. Bad. | |
46 | ||
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. | |
52 | ||
53 | Possibly if we did one directory at a time that would be sufficient. | |
54 | ||
55 | Alternatively we could pre-process the arguments to make sure no | |
58379559 MP |
56 | duplicates will ever be inserted. There could be some bad cases |
57 | when we're collapsing symlinks. | |
b3e6c815 MP |
58 | |
59 | We could have a hash table. | |
60 | ||
a6a3c3df MP |
61 | Memory accounting |
62 | ||
63 | At exit, show how much memory was used for the file list, etc. | |
64 | ||
b3e6c815 MP |
65 | Also we do a wierd exponential-growth allocation in flist.c. I'm |
66 | not sure this makes sense with modern mallocs. At any rate it will | |
67 | make us allocate a huge amount of memory for large file lists. | |
68 | ||
a6a3c3df MP |
69 | Hard-link handling |
70 | ||
71 | At the moment hardlink handling is very expensive, so it's off by | |
72 | default. It does not need to be so. | |
73 | ||
74 | We can rule out hardlinked directories since they will probably | |
75 | screw us up in all kinds of ways. They simply should not be used. | |
76 | ||
77 | At the moment rsync only cares about hardlinks to regular files. I | |
78 | guess you could also use them for sockets, devices and other beasts, | |
79 | but I have not seen them. | |
80 | ||
81 | When trying to reproduce hard links, we only need to worry about | |
82 | files that have more than one name (nlinks>1 && !S_ISDIR). | |
83 | ||
84 | The basic point of this is to discover alternate names that refer to | |
85 | the same file. All operations, including creating the file and | |
86 | writing modifications to it need only to be done for the first name. | |
87 | For all later names, we just create the link and then leave it | |
88 | alone. | |
89 | ||
90 | If hard links are to be preserved: | |
91 | ||
92 | Before the generator/receiver fork, the list of files is received | |
93 | from the sender (recv_file_list), and a table for detecting hard | |
94 | links is built. | |
95 | ||
96 | The generator looks for hard links within the file list and does | |
97 | not send checksums for them, though it does send other metadata. | |
98 | ||
99 | The sender sends the device number and inode with file entries, so | |
100 | that files are uniquely identified. | |
101 | ||
102 | The receiver goes through and creates hard links (do_hard_links) | |
103 | after all data has been written, but before directory permissions | |
104 | are set. | |
105 | ||
106 | At the moment device and inum are sent as 4-byte integers, which | |
107 | will probably cause problems on large filesystems. On Linux the | |
108 | kernel uses 64-bit ino_t's internally, and people will soon have | |
109 | filesystems big enough to use them. We ought to follow NFS4 in | |
110 | using 64-bit device and inode identification, perhaps with a | |
111 | protocol version bump. | |
112 | ||
113 | Once we've seen all the names for a particular file, we no longer | |
114 | need to think about it and we can deallocate the memory. | |
115 | ||
116 | We can also have the case where there are links to a file that are | |
117 | not in the tree being transferred. There's nothing we can do about | |
118 | that. Because we rename the destination into place after writing, | |
119 | any hardlinks to the old file are always going to be orphaned. In | |
120 | fact that is almost necessary because otherwise we'd get really | |
121 | confused if we were generating checksums for one name of a file and | |
122 | modifying another. | |
123 | ||
124 | At the moment the code seems to make a whole second copy of the file | |
125 | list, which seems unnecessary. | |
126 | ||
127 | We should have a test case that exercises hard links. Since it | |
128 | might be hard to compare ./tls output where the inodes change we | |
129 | might need a little program to check whether several names refer to | |
130 | the same file. | |
a2d2e5c0 MP |
131 | |
132 | IPv6 | |
133 | ||
c33e3e39 MP |
134 | Implement suggestions from http://www.kame.net/newsletter/19980604/ |
135 | and ftp://ftp.iij.ad.jp/pub/RFC/rfc2553.txt | |
136 | ||
137 | If a host has multiple addresses, then listen try to connect to all | |
138 | in order until we get through. (getaddrinfo may return multiple | |
c10b0bdd | 139 | addresses.) This is kind of implemented already. |
c33e3e39 MP |
140 | |
141 | Possibly also when starting as a server we may need to listen on | |
142 | multiple passive addresses. This might be a bit harder, because we | |
143 | may need to select on all of them. Hm. | |
144 | ||
a2d2e5c0 MP |
145 | Define a syntax for IPv6 literal addresses. Since they include |
146 | colons, they tend to break most naming systems, including ours. | |
147 | Based on the HTTP IPv6 syntax, I think we should use | |
148 | ||
149 | rsync://[::1]/foo/bar | |
150 | [::1]::bar | |
151 | ||
152 | which should just take a small change to the parser code. | |
153 | ||
5aafd07b MP |
154 | Errors |
155 | ||
156 | If we hang or get SIGINT, then explain where we were up to. Perhaps | |
157 | have a static buffer that contains the current function name, or | |
158 | some kind of description of what we were trying to do. This is a | |
159 | little easier on people than needing to run strace/truss. | |
160 | ||
161 | "The dungeon collapses! You are killed." Rather than "unexpected | |
162 | eof" give a message that is more detailed if possible and also more | |
163 | helpful. | |
164 | ||
5575de14 MP |
165 | File attributes |
166 | ||
167 | Device major/minor numbers should be at least 32 bits each. See | |
168 | http://lists.samba.org/pipermail/rsync/2001-November/005357.html | |
169 | ||
170 | Transfer ACLs. Need to think of a standard representation. | |
171 | Probably better not to even try to convert between NT and POSIX. | |
172 | Possibly can share some code with Samba. | |
5aafd07b | 173 | |
28a69e25 MP |
174 | Empty directories |
175 | ||
176 | With the current common --include '*/' --exclude '*' pattern, people | |
177 | can end up with many empty directories. We might avoid this by | |
178 | lazily creating such directories. | |
179 | ||
180 | zlib | |
181 | ||
182 | Perhaps don't use our own zlib. Will we actually be incompatible, | |
183 | or just be slightly less efficient? | |
184 | ||
185 | logging | |
186 | ||
187 | Perhaps flush stdout after each filename, so that people trying to | |
188 | monitor progress in a log file can do so more easily. See | |
189 | http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=48108 | |
190 | ||
7c583c73 MP |
191 | rsyncd over ssh |
192 | ||
193 | There are already some patches to do this. | |
194 | ||
a2d2e5c0 MP |
195 | PLATFORMS ------------------------------------------------------------ |
196 | ||
197 | Win32 | |
198 | ||
199 | Don't detach, because this messes up --srvany. | |
200 | ||
201 | http://sources.redhat.com/ml/cygwin/2001-08/msg00234.html | |
202 | ||
203 | According to "Effective TCP/IP Programming" (??) close() on a socket | |
204 | has incorrect behaviour on Windows -- it sends a RST packet to the | |
205 | other side, which gives a "connection reset by peer" error. On that | |
206 | platform we should probably do shutdown() instead. However, on Unix | |
207 | we are correct to call close(), because shutdown() discards | |
208 | untransmitted data. | |
209 | ||
7c583c73 MP |
210 | DOCUMENTATION -------------------------------------------------------- |
211 | ||
212 | Update README | |
213 | ||
a2d2e5c0 MP |
214 | BUILD FARM ----------------------------------------------------------- |
215 | ||
216 | Add machines | |
217 | ||
218 | AMDAHL UTS (Dave Dykstra) | |
219 | ||
220 | Cygwin (on different versions of Win32?) | |
221 | ||
222 | HP-UX variants (via HP?) | |
33d213bb | 223 | |
5aafd07b MP |
224 | SCO |
225 | ||
46ef7d1d MP |
226 | NICE ----------------------------------------------------------------- |
227 | ||
71b3374b MP |
228 | SIGHUP |
229 | ||
230 | Re-read config file (just exec() ourselves) rather than exiting. | |
231 | ||
a2d2e5c0 MP |
232 | --no-detach and --no-fork options |
233 | ||
234 | Very useful for debugging. Also good when running under a | |
235 | daemon-monitoring process that tries to restart the service when the | |
236 | parent exits. | |
237 | ||
238 | hang/timeout friendliness | |
239 | ||
50f2f002 MP |
240 | verbose output |
241 | ||
242 | Indicate whether files are new, updated, or deleted | |
243 | ||
a2d2e5c0 MP |
244 | internationalization |
245 | ||
246 | Change to using gettext(). Probably need to ship this for platforms | |
247 | that don't have it. | |
248 | ||
249 | Solicit translations. | |
250 | ||
251 | Does anyone care? | |
252 | ||
46ef7d1d MP |
253 | rsyncsh |
254 | ||
255 | Write a small emulation of interactive ftp as a Pythonn program | |
256 | that calls rsync. Commands such as "cd", "ls", "ls *.c" etc map | |
257 | fairly directly into rsync commands: it just needs to remember the | |
258 | current host, directory and so on. We can probably even do | |
259 | completion of remote filenames. | |
260 | ||
b3e6c815 | 261 | %K% |