3 BUGS ---------------------------------------------------------------
4 Fix progress indicator to not corrupt log
6 Do not rely on having a group called "nobody"
7 Incorrect timestamps (Debian #100295)
10 FEATURES ------------------------------------------------------------
11 server-imposed bandwidth limits
13 Use chroot only if supported
14 Allow supplementary groups in rsyncd.conf 2002/04/09
15 Handling IPv6 on old machines
17 Add ACL support 2001/12/02
18 Lazy directory creation
19 Conditional -z for old protocols
20 proxy authentication 2002/01/23
23 Allow forcing arbitrary permissions 2002/03/12
24 --diff david.e.sewell 2002/03/15
25 Add daemon --no-detach and --no-fork options
26 Create more granular verbosity jw 2003/05/15
28 DOCUMENTATION --------------------------------------------------------
30 Keep list of open issues and todos on the web site
31 Update web site from CVS
32 Perhaps redo manual as SGML
34 LOGGING --------------------------------------------------------------
35 Make dry run list all updates 2002/04/03
37 Improve error messages
38 Better statistics: Rasmus 2002/03/08
39 Perhaps flush stdout like syslog
40 Log deamon sessions that just list modules
41 Log child death on signal
42 Keep stderr and stdout properly separated (Debian #23626)
43 Log errors with function that reports process of origin
44 verbose output David Stein 2001/12/20
45 Add reason for transfer to file logging
46 debugging of daemon 2002/04/08
49 DEVELOPMENT --------------------------------------------------------
50 Handling duplicate names
51 Use generic zlib 2002/02/25
56 Add machines to build farm
58 PERFORMANCE ----------------------------------------------------------
59 File list structure in memory
60 Traverse just one directory at a time
61 Allow skipping MD4 file_sum 2002/04/08
64 TESTING --------------------------------------------------------------
66 Cross-test versions 2001/08/22
69 Create mutator program for testing
70 Create configure option to enable dangerous tests
71 If tests are skipped, say why.
72 Test daemon feature to disallow particular options.
73 Create pipe program for testing
74 Create test makefile target for some tests
75 Test "refuse options" works
77 RELATED PROJECTS -----------------------------------------------------
79 http://rsync.samba.org/rsync-and-debian/
81 rsyncsplit as alternative to real integration with gzip?
82 reverse rsync over HTTP Range
86 BUGS ---------------------------------------------------------------
88 Fix progress indicator to not corrupt log
90 Progress indicator can produce corrupt output when transferring directories:
93 main/binary-arm/admin/
95 main/binary-arm/comm/8.56kB/s 0:00:52
96 main/binary-arm/devel/
98 main/binary-arm/editors/
99 main/binary-arm/electronics/s 0:00:53
100 main/binary-arm/games/
101 main/binary-arm/graphics/
102 main/binary-arm/hamradio/
103 main/binary-arm/interpreters/
104 main/binary-arm/libs/6.61kB/s 0:00:54
105 main/binary-arm/mail/
106 main/binary-arm/math/
107 main/binary-arm/misc/
114 I don't think we handle this properly on systems that don't have the
115 call. Are there any such?
120 Do not rely on having a group called "nobody"
122 http://www.linuxbase.org/spec/refspecs/LSB_1.1.0/gLSB/usernames.html
124 On Debian it's "nogroup"
129 Incorrect timestamps (Debian #100295)
131 A bit hard to believe, but apparently it happens.
138 Don't detach, because this messes up --srvany.
140 http://sources.redhat.com/ml/cygwin/2001-08/msg00234.html
146 FEATURES ------------------------------------------------------------
148 server-imposed bandwidth limits
155 There are already some patches to do this.
157 BitKeeper uses a server whose login shell is set to bkd. That's
158 probably a reasonable approach.
163 Use chroot only if supported
165 If the platform doesn't support it, then don't even try.
167 If running as non-root, then don't fail, just give a warning.
168 (There was a thread about this a while ago?)
170 http://lists.samba.org/pipermail/rsync/2001-August/thread.html
171 http://lists.samba.org/pipermail/rsync/2001-September/thread.html
176 Allow supplementary groups in rsyncd.conf 2002/04/09
178 Perhaps allow supplementary groups to be specified in rsyncd.conf;
179 then make the first one the primary gid and all the rest be
185 Handling IPv6 on old machines
187 The KAME IPv6 patch is nice in theory but has proved a bit of a
188 nightmare in practice. The basic idea of their patch is that rsync
189 is rewritten to use the new getaddrinfo()/getnameinfo() interface,
190 rather than gethostbyname()/gethostbyaddr() as in rsync 2.4.6.
191 Systems that don't have the new interface are handled by providing
192 our own implementation in lib/, which is selectively linked in.
194 The problem with this is that it is really hard to get right on
195 platforms that have a half-working implementation, so redefining
196 these functions clashes with system headers, and leaving them out
197 breaks. This affects at least OSF/1, RedHat 5, and Cobalt, which
198 are moderately improtant.
200 Perhaps the simplest solution would be to have two different files
201 implementing the same interface, and choose either the new or the
202 old API. This is probably necessary for systems that e.g. have
203 IPv6, but gethostbyaddr() can't handle it. The Linux manpage claims
204 this is currently the case.
206 In fact, our internal sockets interface (things like
207 open_socket_out(), etc) is much narrower than the getaddrinfo()
208 interface, and so probably simpler to get right. In addition, the
209 old code is known to work well on old machines.
211 We could drop the rather large lib/getaddrinfo files.
218 Implement suggestions from http://www.kame.net/newsletter/19980604/
219 and ftp://ftp.iij.ad.jp/pub/RFC/rfc2553.txt
221 If a host has multiple addresses, then listen try to connect to all
222 in order until we get through. (getaddrinfo may return multiple
223 addresses.) This is kind of implemented already.
225 Possibly also when starting as a server we may need to listen on
226 multiple passive addresses. This might be a bit harder, because we
227 may need to select on all of them. Hm.
229 Define a syntax for IPv6 literal addresses. Since they include
230 colons, they tend to break most naming systems, including ours.
231 Based on the HTTP IPv6 syntax, I think we should use
233 rsync://[::1]/foo/bar [::1]::bar
235 which should just take a small change to the parser code.
240 Add ACL support 2001/12/02
242 Transfer ACLs. Need to think of a standard representation.
243 Probably better not to even try to convert between NT and POSIX.
244 Possibly can share some code with Samba.
249 Lazy directory creation
251 With the current common --include '*/' --exclude '*' pattern, people
252 can end up with many empty directories. We might avoid this by
253 lazily creating such directories.
258 Conditional -z for old protocols
260 After we get the @RSYNCD greeting from the server, we know it's
261 version but we have not yet sent the command line, so we could just
262 remove the -z option if the server is too old.
264 For ssh invocation it's not so simple, because we actually use the
265 command line to start the remote process. However, we only actually
266 do compression in token.c, and we could therefore once we discover
267 the remote version emit an error if it's too old. I'm not sure if
268 that's a good tradeoff or not.
273 proxy authentication 2002/01/23
275 Allow RSYNC_PROXY to be http://user:pass@proxy.foo:3128/, and do
276 HTTP Basic Proxy-Authentication.
278 Multiple schemes are possible, up to and including the insanity that
279 is NTLM, but Basic probably covers most cases.
286 Add --with-socks, and then perhaps a command-line option to put them
287 on or off. This might be more reliable than LD_PRELOAD hacks.
294 rsync to a FAT partition on a Unix machine doesn't work very well at
295 the moment. I think we get errors about invalid filenames and
296 perhaps also trying to do atomic renames.
298 I guess the code to do this is currently #ifdef'd on Windows;
299 perhaps we ought to intelligently fall back to it on Unix too.
304 Allow forcing arbitrary permissions 2002/03/12
306 On 12 Mar 2002, Dave Dykstra <dwd@bell-labs.com> wrote:
307 > If we would add an option to do that functionality, I
308 > would vote for one that was more general which could mask
309 > off any set of permission bits and possibly add any set of
310 > bits. Perhaps a chmod-like syntax if it could be
311 > implemented simply.
313 I think that would be good too. For example, people uploading files
314 to a web server might like to say
316 rsync -avzP --chmod a+rX ./ sourcefrog.net:/home/www/sourcefrog/
318 Ideally the patch would implement as many of the gnu chmod semantics
319 as possible. I think the mode parser should be a separate function
320 that passes back something like (mask,set) description to the rest
321 of the program. For bonus points there would be a test case for the
324 Possibly also --chown
331 --diff david.e.sewell 2002/03/15
333 Allow people to specify the diff command. (Might want to use wdiff,
336 Just diff the temporary file with the destination file, and delete
337 the tmp file rather than moving it into place.
339 Interaction with --partial.
341 Security interactions with daemon mode?
346 Add daemon --no-detach and --no-fork options
348 Very useful for debugging. Also good when running under a
349 daemon-monitoring process that tries to restart the service when the
355 Create more granular verbosity jw 2003/05/15
357 Control output with the --report option.
359 The option takes as a single argument (no whitespace) a
360 comma delimited lists of keywords.
362 This would separate debugging from "logging" as well as
363 fine grained selection of statistical reporting and what
366 http://lists.samba.org/archive/rsync/2003-May/006059.html
370 DOCUMENTATION --------------------------------------------------------
377 Keep list of open issues and todos on the web site
382 Update web site from CVS
387 Perhaps redo manual as SGML
389 The man page is getting rather large, and there is more information
390 that ought to be added.
392 TexInfo source is probably a dying format.
394 Linuxdoc looks like the most likely contender. I know DocBook is
395 favoured by some people, but it's so bloody verbose, even with emacs
400 LOGGING --------------------------------------------------------------
402 Make dry run list all updates 2002/04/03
406 Mark Santcroos points out that -n fails to list files which have
407 only metadata changes, though it probably should.
409 There may be a Debian bug about this as well.
416 At exit, show how much memory was used for the file list, etc.
418 Also we do a wierd exponential-growth allocation in flist.c. I'm
419 not sure this makes sense with modern mallocs. At any rate it will
420 make us allocate a huge amount of memory for large file lists.
425 Improve error messages
427 If we hang or get SIGINT, then explain where we were up to. Perhaps
428 have a static buffer that contains the current function name, or
429 some kind of description of what we were trying to do. This is a
430 little easier on people than needing to run strace/truss.
432 "The dungeon collapses! You are killed." Rather than "unexpected
433 eof" give a message that is more detailed if possible and also more
436 If we get an error writing to a socket, then we should perhaps
437 continue trying to read to see if an error message comes across
438 explaining why the socket is closed. I'm not sure if this would
439 work, but it would certainly make our messages more helpful.
441 What happens if a directory is missing -x attributes. Do we lose
442 our load? (Debian #28416) Probably fixed now, but a test case would
450 Better statistics: Rasmus 2002/03/08
453 hey, how about an rsync option that just gives you the
454 summary without the list of files? And perhaps gives
455 more information like the number of new files, number
456 of changed, deleted, etc. ?
459 nice idea there is --stats but at the moment it's very
460 tridge-oriented rather than user-friendly it would be
461 nice to improve it that would also work well with
467 Perhaps flush stdout like syslog
469 Perhaps flush stdout after each filename, so that people trying to
470 monitor progress in a log file can do so more easily. See
471 http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=48108
476 Log deamon sessions that just list modules
478 At the connections that just get a list of modules are not logged,
484 Log child death on signal
486 If a child of the rsync daemon dies with a signal, we should notice
487 that when we reap it and log a message.
492 Keep stderr and stdout properly separated (Debian #23626)
497 Log errors with function that reports process of origin
499 Use a separate function for reporting errors; prefix it with
500 "rsync:" or "rsync(remote)", or perhaps even "rsync(local
506 verbose output David Stein 2001/12/20
508 Indicate whether files are new, updated, or deleted
510 At end of transfer, show how many files were or were not transferred
516 Add reason for transfer to file logging
518 Explain *why* every file is transferred or not (e.g. "local mtime
519 123123 newer than 1283198")
524 debugging of daemon 2002/04/08
526 Add an rsyncd.conf parameter to turn on debugging on the server.
533 Change to using gettext(). Probably need to ship this for platforms
536 Solicit translations.
538 Does anyone care? Before we bother modifying the code, we ought to
539 get the manual translated first, because that's possibly more useful
540 and at any rate demonstrates desire.
544 DEVELOPMENT --------------------------------------------------------
546 Handling duplicate names
548 We need to be careful of duplicate names getting into the file list.
549 See clean_flist(). This could happen if multiple arguments include
552 I think duplicates are only a problem if they're both flowing
553 through the pipeline at the same time. For example we might have
554 updated the first occurrence after reading the checksums for the
555 second. So possibly we just need to make sure that we don't have
556 both in the pipeline at the same time.
558 Possibly if we did one directory at a time that would be sufficient.
560 Alternatively we could pre-process the arguments to make sure no
561 duplicates will ever be inserted. There could be some bad cases
562 when we're collapsing symlinks.
564 We could have a hash table.
566 The root of the problem is that we do not want more than one file
567 list entry referring to the same file. At first glance there are
568 several ways this could happen: symlinks, hardlinks, and repeated
569 names on the command line.
571 If names are repeated on the command line, they may be present in
572 different forms, perhaps by traversing directory paths in different
573 ways, traversing paths including symlinks. Also we need to allow
574 for expansion of globs by rsync.
576 At the moment, clean_flist() requires having the entire file list in
577 memory. Duplicate names are detected just by a string comparison.
579 We don't need to worry about hard links causing duplicates because
580 files are never updated in place. Similarly for symlinks.
582 I think even if we're using a different symlink mode we don't need
585 Unless we're really clever this will introduce a protocol
586 incompatibility, so we need to be able to accept the old format as
592 Use generic zlib 2002/02/25
594 Perhaps don't use our own zlib.
598 - will automatically be up to date with bugfixes in zlib
600 - can leave it out for small rsync on e.g. recovery disks
602 - can use a shared library
604 - avoids people breaking rsync by trying to do this themselves and
607 Should we ship zlib for systems that don't have it, or require
608 people to install it separately?
610 Apparently this will make us incompatible with versions of rsync
611 that use the patched version of rsync. Probably the simplest way to
612 do this is to just disable gzip (with a warning) when talking to old
620 Rather than storing the file list in memory, store it in a TDB.
622 This *might* make memory usage lower while building the file list.
624 Hashtable lookup will mean files are not transmitted in order,
627 This would neatly eliminate one of the major post-fork shared data
635 Build rsync with SPLINT to try to find security holes. Add
636 annotations as necessary. Keep track of the number of warnings
637 found initially, and see how many of them are real bugs, or real
638 security bugs. Knowing the percentage of likely hits would be
639 really interesting for other projects.
646 jra recommends Valgrind:
648 http://devel-home.kde.org/~sewardj/
653 Create release script
659 Build tar file; upload
661 Send announcement to mailing list and c.o.l.a.
663 Make freshmeat announcement
670 Add machines to build farm
672 Cygwin (on different versions of Win32?)
674 HP-UX variants (via HP?)
682 PERFORMANCE ----------------------------------------------------------
684 File list structure in memory
686 Rather than one big array, perhaps have a tree in memory mirroring
689 This might make sorting much faster! (I'm not sure it's a big CPU
692 It might also reduce memory use in storing repeated directory names
693 -- again I'm not sure this is a problem.
698 Traverse just one directory at a time
700 Traverse just one directory at a time. Tridge says it's possible.
702 At the moment rsync reads the whole file list into memory at the
703 start, which makes us use a lot of memory and also not pipeline
704 network access as much as we could.
709 Allow skipping MD4 file_sum 2002/04/08
711 If we're doing a local transfer, or using -W, then perhaps don't
712 send the file checksum. If we're doing a local transfer, then
713 calculating MD4 checksums uses 90% of CPU and is unlikely to be
716 Indeed for transfers over zlib or ssh we can also rely on the
717 transport to have quite strong protection against corruption.
719 Perhaps we should have an option to disable this,
720 analogous to --whole-file, although it would default to
721 disabled. The file checksum takes up a definite space in
722 the protocol -- we can either set it to 0, or perhaps just
730 Perhaps borrow an assembler MD4 from someone?
732 Make sure we call MD4 with properly-sized blocks whenever possible
733 to avoid copying into the residue region?
737 TESTING --------------------------------------------------------------
741 Something that just keeps running rsync continuously over a data set
742 likely to generate problems.
747 Cross-test versions 2001/08/22
749 Part of the regression suite should be making sure that we
750 don't break backwards compatibility: old clients vs new
751 servers and so on. Ideally we would test both up and down
752 from the current release to all old versions.
754 Run current rsync versions against significant past releases.
756 We might need to omit broken old versions, or versions in which
757 particular functionality is broken
759 It might be sufficient to test downloads from well-known public
760 rsync servers running different versions of rsync. This will give
761 some testing and also be the most common case for having different
762 versions and not being able to upgrade.
764 The new --protocol option may help in this.
769 Test on kernel source
771 Download all versions of kernel; unpack, sync between them. Also
772 sync between uncompressed tarballs. Compare directories after
775 Use local mode; ssh; daemon; --whole-file and --no-whole-file.
777 Use awk to pull out the 'speedup' number for each transfer. Make
785 Sparse and non-sparse
790 Create mutator program for testing
792 Insert bytes, delete bytes, swap blocks, ...
797 Create configure option to enable dangerous tests
802 If tests are skipped, say why.
807 Test daemon feature to disallow particular options.
812 Create pipe program for testing
814 Create pipe program that makes slow/jerky connections for
815 testing Versions of read() and write() that corrupt the
816 stream, or abruptly fail
821 Create test makefile target for some tests
823 Separate makefile target to run rough tests -- or perhaps
824 just run them every time?
829 Test "refuse options" works
831 What about for --recursive?
833 If you specify an unrecognized option here, you should get an error.
835 We need a test case for this...
837 Was this broken when we changed to popt?
841 RELATED PROJECTS -----------------------------------------------------
845 Write a small emulation of interactive ftp as a Pythonn program
846 that calls rsync. Commands such as "cd", "ls", "ls *.c" etc map
847 fairly directly into rsync commands: it just needs to remember the
848 current host, directory and so on. We can probably even do
849 completion of remote filenames.
854 http://rsync.samba.org/rsync-and-debian/
862 Exhaustive, tortuous testing
869 rsyncsplit as alternative to real integration with gzip?
874 reverse rsync over HTTP Range
876 Goswin Brederlow suggested this on Debian; I think tridge and I
877 talked about it previous in relation to rproxy.