3 FEATURES ------------------------------------------------------------
4 Use chroot only if supported
5 Allow supplementary groups in rsyncd.conf 2002/04/09
6 Handling IPv6 on old machines
8 Add ACL support 2001/12/02
9 proxy authentication 2002/01/23
12 --diff david.e.sewell 2002/03/15
13 Add daemon --no-fork option
14 Create more granular verbosity 2003/05/15
16 DOCUMENTATION --------------------------------------------------------
17 Keep list of open issues and todos on the web site
18 Perhaps redo manual as SGML
20 LOGGING --------------------------------------------------------------
22 Improve error messages
23 Better statistics Rasmus 2002/03/08
24 Perhaps flush stdout like syslog
25 Log child death on signal
26 verbose output David Stein 2001/12/20
29 DEVELOPMENT --------------------------------------------------------
30 Handling duplicate names
31 Use generic zlib 2002/02/25
35 PERFORMANCE ----------------------------------------------------------
36 Traverse just one directory at a time
37 Allow skipping MD4 file_sum 2002/04/08
40 TESTING --------------------------------------------------------------
42 Cross-test versions 2001/08/22
45 Create mutator program for testing
46 Create configure option to enable dangerous tests
47 Create pipe program for testing
48 Create test makefile target for some tests
50 RELATED PROJECTS -----------------------------------------------------
52 http://rsync.samba.org/rsync-and-debian/
54 rsyncsplit as alternative to real integration with gzip?
55 reverse rsync over HTTP Range
59 FEATURES ------------------------------------------------------------
62 Use chroot only if supported
64 If the platform doesn't support it, then don't even try.
66 If running as non-root, then don't fail, just give a warning.
67 (There was a thread about this a while ago?)
69 http://lists.samba.org/pipermail/rsync/2001-August/thread.html
70 http://lists.samba.org/pipermail/rsync/2001-September/thread.html
75 Allow supplementary groups in rsyncd.conf 2002/04/09
77 Perhaps allow supplementary groups to be specified in rsyncd.conf;
78 then make the first one the primary gid and all the rest be
84 Handling IPv6 on old machines
86 The KAME IPv6 patch is nice in theory but has proved a bit of a
87 nightmare in practice. The basic idea of their patch is that rsync
88 is rewritten to use the new getaddrinfo()/getnameinfo() interface,
89 rather than gethostbyname()/gethostbyaddr() as in rsync 2.4.6.
90 Systems that don't have the new interface are handled by providing
91 our own implementation in lib/, which is selectively linked in.
93 The problem with this is that it is really hard to get right on
94 platforms that have a half-working implementation, so redefining
95 these functions clashes with system headers, and leaving them out
96 breaks. This affects at least OSF/1, RedHat 5, and Cobalt, which
97 are moderately improtant.
99 Perhaps the simplest solution would be to have two different files
100 implementing the same interface, and choose either the new or the
101 old API. This is probably necessary for systems that e.g. have
102 IPv6, but gethostbyaddr() can't handle it. The Linux manpage claims
103 this is currently the case.
105 In fact, our internal sockets interface (things like
106 open_socket_out(), etc) is much narrower than the getaddrinfo()
107 interface, and so probably simpler to get right. In addition, the
108 old code is known to work well on old machines.
110 We could drop the rather large lib/getaddrinfo files.
117 Implement suggestions from http://www.kame.net/newsletter/19980604/
118 and ftp://ftp.iij.ad.jp/pub/RFC/rfc2553.txt
120 If a host has multiple addresses, then listen try to connect to all
121 in order until we get through. (getaddrinfo may return multiple
122 addresses.) This is kind of implemented already.
124 Possibly also when starting as a server we may need to listen on
125 multiple passive addresses. This might be a bit harder, because we
126 may need to select on all of them. Hm.
131 Add ACL support 2001/12/02
133 Transfer ACLs. Need to think of a standard representation.
134 Probably better not to even try to convert between NT and POSIX.
135 Possibly can share some code with Samba.
136 NOTE: there is a patch that implements this in the "patches" subdir.
141 proxy authentication 2002/01/23
143 Allow RSYNC_PROXY to be http://user:pass@proxy.foo:3128/, and do
144 HTTP Basic Proxy-Authentication.
146 Multiple schemes are possible, up to and including the insanity that
147 is NTLM, but Basic probably covers most cases.
154 Add --with-socks, and then perhaps a command-line option to put them
155 on or off. This might be more reliable than LD_PRELOAD hacks.
162 rsync to a FAT partition on a Unix machine doesn't work very well at
163 the moment. I think we get errors about invalid filenames and
164 perhaps also trying to do atomic renames.
166 I guess the code to do this is currently #ifdef'd on Windows;
167 perhaps we ought to intelligently fall back to it on Unix too.
172 --diff david.e.sewell 2002/03/15
174 Allow people to specify the diff command. (Might want to use wdiff,
177 Just diff the temporary file with the destination file, and delete
178 the tmp file rather than moving it into place.
180 Interaction with --partial.
182 Security interactions with daemon mode?
187 Add daemon --no-fork option
189 Very useful for debugging. Also good when running under a
190 daemon-monitoring process that tries to restart the service when the
196 Create more granular verbosity 2003/05/15
198 Control output with the --report option.
200 The option takes as a single argument (no whitespace) a
201 comma delimited lists of keywords.
203 This would separate debugging from "logging" as well as
204 fine grained selection of statistical reporting and what
207 http://lists.samba.org/archive/rsync/2003-May/006059.html
211 DOCUMENTATION --------------------------------------------------------
214 Keep list of open issues and todos on the web site
219 Perhaps redo manual as SGML
221 The man page is getting rather large, and there is more information
222 that ought to be added.
224 TexInfo source is probably a dying format.
226 Linuxdoc looks like the most likely contender. I know DocBook is
227 favoured by some people, but it's so bloody verbose, even with emacs
232 LOGGING --------------------------------------------------------------
237 At exit, show how much memory was used for the file list, etc.
239 Also we do a wierd exponential-growth allocation in flist.c. I'm
240 not sure this makes sense with modern mallocs. At any rate it will
241 make us allocate a huge amount of memory for large file lists.
246 Improve error messages
248 If we hang or get SIGINT, then explain where we were up to. Perhaps
249 have a static buffer that contains the current function name, or
250 some kind of description of what we were trying to do. This is a
251 little easier on people than needing to run strace/truss.
253 "The dungeon collapses! You are killed." Rather than "unexpected
254 eof" give a message that is more detailed if possible and also more
257 If we get an error writing to a socket, then we should perhaps
258 continue trying to read to see if an error message comes across
259 explaining why the socket is closed. I'm not sure if this would
260 work, but it would certainly make our messages more helpful.
262 What happens if a directory is missing -x attributes. Do we lose
263 our load? (Debian #28416) Probably fixed now, but a test case would
269 Better statistics Rasmus 2002/03/08
272 hey, how about an rsync option that just gives you the
273 summary without the list of files? And perhaps gives
274 more information like the number of new files, number
275 of changed, deleted, etc. ?
278 nice idea there is --stats but at the moment it's very
279 tridge-oriented rather than user-friendly it would be
280 nice to improve it that would also work well with
286 Perhaps flush stdout like syslog
288 Perhaps flush stdout after each filename, so that people trying to
289 monitor progress in a log file can do so more easily. See
290 http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=48108
295 Log child death on signal
297 If a child of the rsync daemon dies with a signal, we should notice
298 that when we reap it and log a message.
303 verbose output David Stein 2001/12/20
305 At end of transfer, show how many files were or were not transferred
313 Change to using gettext(). Probably need to ship this for platforms
316 Solicit translations.
318 Does anyone care? Before we bother modifying the code, we ought to
319 get the manual translated first, because that's possibly more useful
320 and at any rate demonstrates desire.
324 DEVELOPMENT --------------------------------------------------------
326 Handling duplicate names
328 Some folks would like rsync to be deterministic in how it handles
329 duplicate names that come from mering multiple source directories
330 into a single destination directory; e.g. the last name wins. We
331 could do this by switching our sort algorithm to one that will
332 guarantee that the names won't be reordered. Alternately, we could
333 assign an ever-increasing number to each item as we insert it into
334 the list and then make sure that we leave the largest number when
335 cleaning the file list (see clean_flist()). Another solution would
336 be to add a hash table, and thus never put any duplicate names into
337 the file list (and bump the protocol to handle this).
342 Use generic zlib 2002/02/25
344 Perhaps don't use our own zlib.
348 - will automatically be up to date with bugfixes in zlib
350 - can leave it out for small rsync on e.g. recovery disks
352 - can use a shared library
354 - avoids people breaking rsync by trying to do this themselves and
357 Should we ship zlib for systems that don't have it, or require
358 people to install it separately?
360 Apparently this will make us incompatible with versions of rsync
361 that use the patched version of rsync. Probably the simplest way to
362 do this is to just disable gzip (with a warning) when talking to old
370 Rather than storing the file list in memory, store it in a TDB.
372 This *might* make memory usage lower while building the file list.
374 Hashtable lookup will mean files are not transmitted in order,
377 This would neatly eliminate one of the major post-fork shared data
385 Build rsync with SPLINT to try to find security holes. Add
386 annotations as necessary. Keep track of the number of warnings
387 found initially, and see how many of them are real bugs, or real
388 security bugs. Knowing the percentage of likely hits would be
389 really interesting for other projects.
393 PERFORMANCE ----------------------------------------------------------
395 Traverse just one directory at a time
397 Traverse just one directory at a time. Tridge says it's possible.
399 At the moment rsync reads the whole file list into memory at the
400 start, which makes us use a lot of memory and also not pipeline
401 network access as much as we could.
406 Allow skipping MD4 file_sum 2002/04/08
408 If we're doing a local transfer, or using -W, then perhaps don't
409 send the file checksum. If we're doing a local transfer, then
410 calculating MD4 checksums uses 90% of CPU and is unlikely to be
413 We should not allow it to be disabled separately from -W, though
414 as it is the only thing that lets us know when the rsync algorithm
415 got out of sync and messed the file up (i.e. if the basis file
416 changed between checksum generation and reception).
423 Perhaps borrow an assembler MD4 from someone?
425 Make sure we call MD4 with properly-sized blocks whenever possible
426 to avoid copying into the residue region?
430 TESTING --------------------------------------------------------------
434 Something that just keeps running rsync continuously over a data set
435 likely to generate problems.
440 Cross-test versions 2001/08/22
442 Part of the regression suite should be making sure that we
443 don't break backwards compatibility: old clients vs new
444 servers and so on. Ideally we would test both up and down
445 from the current release to all old versions.
447 Run current rsync versions against significant past releases.
449 We might need to omit broken old versions, or versions in which
450 particular functionality is broken
452 It might be sufficient to test downloads from well-known public
453 rsync servers running different versions of rsync. This will give
454 some testing and also be the most common case for having different
455 versions and not being able to upgrade.
457 The new --protocol option may help in this.
462 Test on kernel source
464 Download all versions of kernel; unpack, sync between them. Also
465 sync between uncompressed tarballs. Compare directories after
468 Use local mode; ssh; daemon; --whole-file and --no-whole-file.
470 Use awk to pull out the 'speedup' number for each transfer. Make
478 Sparse and non-sparse
483 Create mutator program for testing
485 Insert bytes, delete bytes, swap blocks, ...
490 Create configure option to enable dangerous tests
495 Create pipe program for testing
497 Create pipe program that makes slow/jerky connections for
498 testing Versions of read() and write() that corrupt the
499 stream, or abruptly fail
504 Create test makefile target for some tests
506 Separate makefile target to run rough tests -- or perhaps
507 just run them every time?
511 RELATED PROJECTS -----------------------------------------------------
515 Write a small emulation of interactive ftp as a Pythonn program
516 that calls rsync. Commands such as "cd", "ls", "ls *.c" etc map
517 fairly directly into rsync commands: it just needs to remember the
518 current host, directory and so on. We can probably even do
519 completion of remote filenames.
524 http://rsync.samba.org/rsync-and-debian/
532 Exhaustive, tortuous testing
539 rsyncsplit as alternative to real integration with gzip?
544 reverse rsync over HTTP Range
546 Goswin Brederlow suggested this on Debian; I think tridge and I
547 talked about it previous in relation to rproxy.
549 Addendum: It looks like someone is working on a version of this:
551 http://zsync.moria.org.uk/