3 BUGS ---------------------------------------------------------------
4 Do not rely on having a group called "nobody"
6 FEATURES ------------------------------------------------------------
7 Use chroot only if supported
8 Allow supplementary groups in rsyncd.conf 2002/04/09
9 Handling IPv6 on old machines
11 Add ACL support 2001/12/02
12 Lazy directory creation
13 proxy authentication 2002/01/23
16 Allow forcing arbitrary permissions 2002/03/12
17 --diff david.e.sewell 2002/03/15
18 Add daemon --no-fork option
19 Create more granular verbosity jw 2003/05/15
21 DOCUMENTATION --------------------------------------------------------
22 Keep list of open issues and todos on the web site
23 Perhaps redo manual as SGML
25 LOGGING --------------------------------------------------------------
27 Improve error messages
28 Better statistics: Rasmus 2002/03/08
29 Perhaps flush stdout like syslog
30 Log deamon sessions that just list modules
31 Log child death on signal
32 Log errors with function that reports process of origin
33 verbose output David Stein 2001/12/20
36 DEVELOPMENT --------------------------------------------------------
37 Handling duplicate names
38 Use generic zlib 2002/02/25
42 PERFORMANCE ----------------------------------------------------------
43 File list structure in memory
44 Traverse just one directory at a time
45 Allow skipping MD4 file_sum 2002/04/08
48 TESTING --------------------------------------------------------------
50 Cross-test versions 2001/08/22
53 Create mutator program for testing
54 Create configure option to enable dangerous tests
55 If tests are skipped, say why.
56 Test daemon feature to disallow particular options.
57 Create pipe program for testing
58 Create test makefile target for some tests
60 RELATED PROJECTS -----------------------------------------------------
62 http://rsync.samba.org/rsync-and-debian/
64 rsyncsplit as alternative to real integration with gzip?
65 reverse rsync over HTTP Range
69 BUGS ---------------------------------------------------------------
72 Do not rely on having a group called "nobody"
74 http://www.linuxbase.org/spec/refspecs/LSB_1.1.0/gLSB/usernames.html
76 On Debian it's "nogroup"
80 FEATURES ------------------------------------------------------------
83 Use chroot only if supported
85 If the platform doesn't support it, then don't even try.
87 If running as non-root, then don't fail, just give a warning.
88 (There was a thread about this a while ago?)
90 http://lists.samba.org/pipermail/rsync/2001-August/thread.html
91 http://lists.samba.org/pipermail/rsync/2001-September/thread.html
96 Allow supplementary groups in rsyncd.conf 2002/04/09
98 Perhaps allow supplementary groups to be specified in rsyncd.conf;
99 then make the first one the primary gid and all the rest be
105 Handling IPv6 on old machines
107 The KAME IPv6 patch is nice in theory but has proved a bit of a
108 nightmare in practice. The basic idea of their patch is that rsync
109 is rewritten to use the new getaddrinfo()/getnameinfo() interface,
110 rather than gethostbyname()/gethostbyaddr() as in rsync 2.4.6.
111 Systems that don't have the new interface are handled by providing
112 our own implementation in lib/, which is selectively linked in.
114 The problem with this is that it is really hard to get right on
115 platforms that have a half-working implementation, so redefining
116 these functions clashes with system headers, and leaving them out
117 breaks. This affects at least OSF/1, RedHat 5, and Cobalt, which
118 are moderately improtant.
120 Perhaps the simplest solution would be to have two different files
121 implementing the same interface, and choose either the new or the
122 old API. This is probably necessary for systems that e.g. have
123 IPv6, but gethostbyaddr() can't handle it. The Linux manpage claims
124 this is currently the case.
126 In fact, our internal sockets interface (things like
127 open_socket_out(), etc) is much narrower than the getaddrinfo()
128 interface, and so probably simpler to get right. In addition, the
129 old code is known to work well on old machines.
131 We could drop the rather large lib/getaddrinfo files.
138 Implement suggestions from http://www.kame.net/newsletter/19980604/
139 and ftp://ftp.iij.ad.jp/pub/RFC/rfc2553.txt
141 If a host has multiple addresses, then listen try to connect to all
142 in order until we get through. (getaddrinfo may return multiple
143 addresses.) This is kind of implemented already.
145 Possibly also when starting as a server we may need to listen on
146 multiple passive addresses. This might be a bit harder, because we
147 may need to select on all of them. Hm.
149 Define a syntax for IPv6 literal addresses. Since they include
150 colons, they tend to break most naming systems, including ours.
151 Based on the HTTP IPv6 syntax, I think we should use
153 rsync://[::1]/foo/bar [::1]::bar
155 which should just take a small change to the parser code.
160 Add ACL support 2001/12/02
162 Transfer ACLs. Need to think of a standard representation.
163 Probably better not to even try to convert between NT and POSIX.
164 Possibly can share some code with Samba.
165 NOTE: there is a patch that implements this in the "patches" subdir.
170 Lazy directory creation
172 With the current common --include '*/' --exclude '*' pattern, people
173 can end up with many empty directories. We might avoid this by
174 lazily creating such directories.
179 proxy authentication 2002/01/23
181 Allow RSYNC_PROXY to be http://user:pass@proxy.foo:3128/, and do
182 HTTP Basic Proxy-Authentication.
184 Multiple schemes are possible, up to and including the insanity that
185 is NTLM, but Basic probably covers most cases.
192 Add --with-socks, and then perhaps a command-line option to put them
193 on or off. This might be more reliable than LD_PRELOAD hacks.
200 rsync to a FAT partition on a Unix machine doesn't work very well at
201 the moment. I think we get errors about invalid filenames and
202 perhaps also trying to do atomic renames.
204 I guess the code to do this is currently #ifdef'd on Windows;
205 perhaps we ought to intelligently fall back to it on Unix too.
210 Allow forcing arbitrary permissions 2002/03/12
212 On 12 Mar 2002, Dave Dykstra <dwd@bell-labs.com> wrote:
213 > If we would add an option to do that functionality, I
214 > would vote for one that was more general which could mask
215 > off any set of permission bits and possibly add any set of
216 > bits. Perhaps a chmod-like syntax if it could be
217 > implemented simply.
219 I think that would be good too. For example, people uploading files
220 to a web server might like to say
222 rsync -avzP --chmod a+rX ./ sourcefrog.net:/home/www/sourcefrog/
224 Ideally the patch would implement as many of the gnu chmod semantics
225 as possible. I think the mode parser should be a separate function
226 that passes back something like (mask,set) description to the rest
227 of the program. For bonus points there would be a test case for the
230 Possibly also --chown
234 NOTE: there is a patch that implements this in the "patches" subdir.
239 --diff david.e.sewell 2002/03/15
241 Allow people to specify the diff command. (Might want to use wdiff,
244 Just diff the temporary file with the destination file, and delete
245 the tmp file rather than moving it into place.
247 Interaction with --partial.
249 Security interactions with daemon mode?
254 Add daemon --no-fork option
256 Very useful for debugging. Also good when running under a
257 daemon-monitoring process that tries to restart the service when the
263 Create more granular verbosity jw 2003/05/15
265 Control output with the --report option.
267 The option takes as a single argument (no whitespace) a
268 comma delimited lists of keywords.
270 This would separate debugging from "logging" as well as
271 fine grained selection of statistical reporting and what
274 http://lists.samba.org/archive/rsync/2003-May/006059.html
278 DOCUMENTATION --------------------------------------------------------
281 Keep list of open issues and todos on the web site
286 Perhaps redo manual as SGML
288 The man page is getting rather large, and there is more information
289 that ought to be added.
291 TexInfo source is probably a dying format.
293 Linuxdoc looks like the most likely contender. I know DocBook is
294 favoured by some people, but it's so bloody verbose, even with emacs
299 LOGGING --------------------------------------------------------------
304 At exit, show how much memory was used for the file list, etc.
306 Also we do a wierd exponential-growth allocation in flist.c. I'm
307 not sure this makes sense with modern mallocs. At any rate it will
308 make us allocate a huge amount of memory for large file lists.
313 Improve error messages
315 If we hang or get SIGINT, then explain where we were up to. Perhaps
316 have a static buffer that contains the current function name, or
317 some kind of description of what we were trying to do. This is a
318 little easier on people than needing to run strace/truss.
320 "The dungeon collapses! You are killed." Rather than "unexpected
321 eof" give a message that is more detailed if possible and also more
324 If we get an error writing to a socket, then we should perhaps
325 continue trying to read to see if an error message comes across
326 explaining why the socket is closed. I'm not sure if this would
327 work, but it would certainly make our messages more helpful.
329 What happens if a directory is missing -x attributes. Do we lose
330 our load? (Debian #28416) Probably fixed now, but a test case would
333 When running as a daemon, some errors should both be returned to the
334 user and logged. This will make interacting with a daemon less
340 Better statistics: Rasmus 2002/03/08
343 hey, how about an rsync option that just gives you the
344 summary without the list of files? And perhaps gives
345 more information like the number of new files, number
346 of changed, deleted, etc. ?
349 nice idea there is --stats but at the moment it's very
350 tridge-oriented rather than user-friendly it would be
351 nice to improve it that would also work well with
357 Perhaps flush stdout like syslog
359 Perhaps flush stdout after each filename, so that people trying to
360 monitor progress in a log file can do so more easily. See
361 http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=48108
366 Log deamon sessions that just list modules
368 At the connections that just get a list of modules are not logged,
374 Log child death on signal
376 If a child of the rsync daemon dies with a signal, we should notice
377 that when we reap it and log a message.
382 Log errors with function that reports process of origin
384 Use a separate function for reporting errors; prefix it with
385 "rsync:" or "rsync(remote)", or perhaps even "rsync(local
391 verbose output David Stein 2001/12/20
393 At end of transfer, show how many files were or were not transferred
401 Change to using gettext(). Probably need to ship this for platforms
404 Solicit translations.
406 Does anyone care? Before we bother modifying the code, we ought to
407 get the manual translated first, because that's possibly more useful
408 and at any rate demonstrates desire.
412 DEVELOPMENT --------------------------------------------------------
414 Handling duplicate names
416 Some folks would like rsync to be deterministic in how it handles
417 duplicate names that come from mering multiple source directories
418 into a single destination directory; e.g. the last name wins. We
419 could do this by switching our sort algorithm to one that will
420 guarantee that the names won't be reordered. Alternately, we could
421 assign an ever-increasing number to each item as we insert it into
422 the list and then make sure that we leave the largest number when
423 cleaning the file list (see clean_flist()). Another solution would
424 be to add a hash table, and thus never put any duplicate names into
425 the file list (and bump the protocol to handle this).
430 Use generic zlib 2002/02/25
432 Perhaps don't use our own zlib.
436 - will automatically be up to date with bugfixes in zlib
438 - can leave it out for small rsync on e.g. recovery disks
440 - can use a shared library
442 - avoids people breaking rsync by trying to do this themselves and
445 Should we ship zlib for systems that don't have it, or require
446 people to install it separately?
448 Apparently this will make us incompatible with versions of rsync
449 that use the patched version of rsync. Probably the simplest way to
450 do this is to just disable gzip (with a warning) when talking to old
458 Rather than storing the file list in memory, store it in a TDB.
460 This *might* make memory usage lower while building the file list.
462 Hashtable lookup will mean files are not transmitted in order,
465 This would neatly eliminate one of the major post-fork shared data
473 Build rsync with SPLINT to try to find security holes. Add
474 annotations as necessary. Keep track of the number of warnings
475 found initially, and see how many of them are real bugs, or real
476 security bugs. Knowing the percentage of likely hits would be
477 really interesting for other projects.
481 PERFORMANCE ----------------------------------------------------------
483 File list structure in memory
485 Rather than one big array, perhaps have a tree in memory mirroring
488 This might make sorting much faster! (I'm not sure it's a big CPU
491 It might also reduce memory use in storing repeated directory names
492 -- again I'm not sure this is a problem.
497 Traverse just one directory at a time
499 Traverse just one directory at a time. Tridge says it's possible.
501 At the moment rsync reads the whole file list into memory at the
502 start, which makes us use a lot of memory and also not pipeline
503 network access as much as we could.
508 Allow skipping MD4 file_sum 2002/04/08
510 If we're doing a local transfer, or using -W, then perhaps don't
511 send the file checksum. If we're doing a local transfer, then
512 calculating MD4 checksums uses 90% of CPU and is unlikely to be
515 We should not allow it to be disabled separately from -W, though
516 as it is the only thing that lets us know when the rsync algorithm
517 got out of sync and messed the file up (i.e. if the basis file
518 changed between checksum generation and reception).
525 Perhaps borrow an assembler MD4 from someone?
527 Make sure we call MD4 with properly-sized blocks whenever possible
528 to avoid copying into the residue region?
532 TESTING --------------------------------------------------------------
536 Something that just keeps running rsync continuously over a data set
537 likely to generate problems.
542 Cross-test versions 2001/08/22
544 Part of the regression suite should be making sure that we
545 don't break backwards compatibility: old clients vs new
546 servers and so on. Ideally we would test both up and down
547 from the current release to all old versions.
549 Run current rsync versions against significant past releases.
551 We might need to omit broken old versions, or versions in which
552 particular functionality is broken
554 It might be sufficient to test downloads from well-known public
555 rsync servers running different versions of rsync. This will give
556 some testing and also be the most common case for having different
557 versions and not being able to upgrade.
559 The new --protocol option may help in this.
564 Test on kernel source
566 Download all versions of kernel; unpack, sync between them. Also
567 sync between uncompressed tarballs. Compare directories after
570 Use local mode; ssh; daemon; --whole-file and --no-whole-file.
572 Use awk to pull out the 'speedup' number for each transfer. Make
580 Sparse and non-sparse
585 Create mutator program for testing
587 Insert bytes, delete bytes, swap blocks, ...
592 Create configure option to enable dangerous tests
597 If tests are skipped, say why.
602 Test daemon feature to disallow particular options.
607 Create pipe program for testing
609 Create pipe program that makes slow/jerky connections for
610 testing Versions of read() and write() that corrupt the
611 stream, or abruptly fail
616 Create test makefile target for some tests
618 Separate makefile target to run rough tests -- or perhaps
619 just run them every time?
623 RELATED PROJECTS -----------------------------------------------------
627 Write a small emulation of interactive ftp as a Pythonn program
628 that calls rsync. Commands such as "cd", "ls", "ls *.c" etc map
629 fairly directly into rsync commands: it just needs to remember the
630 current host, directory and so on. We can probably even do
631 completion of remote filenames.
636 http://rsync.samba.org/rsync-and-debian/
644 Exhaustive, tortuous testing
651 rsyncsplit as alternative to real integration with gzip?
656 reverse rsync over HTTP Range
658 Goswin Brederlow suggested this on Debian; I think tridge and I
659 talked about it previous in relation to rproxy.
661 Addendum: It looks like someone is working on a version of this:
663 http://zsync.moria.org.uk/