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 proxy authentication 2002/01/23
22 Allow forcing arbitrary permissions 2002/03/12
23 --diff david.e.sewell 2002/03/15
24 Add daemon --no-detach and --no-fork options
25 Create more granular verbosity jw 2003/05/15
27 DOCUMENTATION --------------------------------------------------------
29 Keep list of open issues and todos on the web site
30 Update web site from CVS
31 Perhaps redo manual as SGML
33 LOGGING --------------------------------------------------------------
34 Make dry run list all updates 2002/04/03
36 Improve error messages
37 Better statistics: Rasmus 2002/03/08
38 Perhaps flush stdout like syslog
39 Log deamon sessions that just list modules
40 Log child death on signal
41 Keep stderr and stdout properly separated (Debian #23626)
42 Log errors with function that reports process of origin
43 verbose output David Stein 2001/12/20
44 Add reason for transfer to file logging
45 debugging of daemon 2002/04/08
48 DEVELOPMENT --------------------------------------------------------
49 Handling duplicate names
50 Use generic zlib 2002/02/25
55 Add machines to build farm
57 PERFORMANCE ----------------------------------------------------------
58 File list structure in memory
59 Traverse just one directory at a time
60 Allow skipping MD4 file_sum 2002/04/08
63 TESTING --------------------------------------------------------------
65 Cross-test versions 2001/08/22
68 Create mutator program for testing
69 Create configure option to enable dangerous tests
70 If tests are skipped, say why.
71 Test daemon feature to disallow particular options.
72 Create pipe program for testing
73 Create test makefile target for some tests
74 Test "refuse options" works
76 RELATED PROJECTS -----------------------------------------------------
78 http://rsync.samba.org/rsync-and-debian/
80 rsyncsplit as alternative to real integration with gzip?
81 reverse rsync over HTTP Range
85 BUGS ---------------------------------------------------------------
87 Fix progress indicator to not corrupt log
89 Progress indicator can produce corrupt output when transferring directories:
92 main/binary-arm/admin/
94 main/binary-arm/comm/8.56kB/s 0:00:52
95 main/binary-arm/devel/
97 main/binary-arm/editors/
98 main/binary-arm/electronics/s 0:00:53
99 main/binary-arm/games/
100 main/binary-arm/graphics/
101 main/binary-arm/hamradio/
102 main/binary-arm/interpreters/
103 main/binary-arm/libs/6.61kB/s 0:00:54
104 main/binary-arm/mail/
105 main/binary-arm/math/
106 main/binary-arm/misc/
113 I don't think we handle this properly on systems that don't have the
114 call. Are there any such?
119 Do not rely on having a group called "nobody"
121 http://www.linuxbase.org/spec/refspecs/LSB_1.1.0/gLSB/usernames.html
123 On Debian it's "nogroup"
128 Incorrect timestamps (Debian #100295)
130 A bit hard to believe, but apparently it happens.
137 Don't detach, because this messes up --srvany.
139 http://sources.redhat.com/ml/cygwin/2001-08/msg00234.html
145 FEATURES ------------------------------------------------------------
147 server-imposed bandwidth limits
154 There are already some patches to do this.
156 BitKeeper uses a server whose login shell is set to bkd. That's
157 probably a reasonable approach.
162 Use chroot only if supported
164 If the platform doesn't support it, then don't even try.
166 If running as non-root, then don't fail, just give a warning.
167 (There was a thread about this a while ago?)
169 http://lists.samba.org/pipermail/rsync/2001-August/thread.html
170 http://lists.samba.org/pipermail/rsync/2001-September/thread.html
175 Allow supplementary groups in rsyncd.conf 2002/04/09
177 Perhaps allow supplementary groups to be specified in rsyncd.conf;
178 then make the first one the primary gid and all the rest be
184 Handling IPv6 on old machines
186 The KAME IPv6 patch is nice in theory but has proved a bit of a
187 nightmare in practice. The basic idea of their patch is that rsync
188 is rewritten to use the new getaddrinfo()/getnameinfo() interface,
189 rather than gethostbyname()/gethostbyaddr() as in rsync 2.4.6.
190 Systems that don't have the new interface are handled by providing
191 our own implementation in lib/, which is selectively linked in.
193 The problem with this is that it is really hard to get right on
194 platforms that have a half-working implementation, so redefining
195 these functions clashes with system headers, and leaving them out
196 breaks. This affects at least OSF/1, RedHat 5, and Cobalt, which
197 are moderately improtant.
199 Perhaps the simplest solution would be to have two different files
200 implementing the same interface, and choose either the new or the
201 old API. This is probably necessary for systems that e.g. have
202 IPv6, but gethostbyaddr() can't handle it. The Linux manpage claims
203 this is currently the case.
205 In fact, our internal sockets interface (things like
206 open_socket_out(), etc) is much narrower than the getaddrinfo()
207 interface, and so probably simpler to get right. In addition, the
208 old code is known to work well on old machines.
210 We could drop the rather large lib/getaddrinfo files.
217 Implement suggestions from http://www.kame.net/newsletter/19980604/
218 and ftp://ftp.iij.ad.jp/pub/RFC/rfc2553.txt
220 If a host has multiple addresses, then listen try to connect to all
221 in order until we get through. (getaddrinfo may return multiple
222 addresses.) This is kind of implemented already.
224 Possibly also when starting as a server we may need to listen on
225 multiple passive addresses. This might be a bit harder, because we
226 may need to select on all of them. Hm.
228 Define a syntax for IPv6 literal addresses. Since they include
229 colons, they tend to break most naming systems, including ours.
230 Based on the HTTP IPv6 syntax, I think we should use
232 rsync://[::1]/foo/bar [::1]::bar
234 which should just take a small change to the parser code.
239 Add ACL support 2001/12/02
241 Transfer ACLs. Need to think of a standard representation.
242 Probably better not to even try to convert between NT and POSIX.
243 Possibly can share some code with Samba.
248 Lazy directory creation
250 With the current common --include '*/' --exclude '*' pattern, people
251 can end up with many empty directories. We might avoid this by
252 lazily creating such directories.
257 proxy authentication 2002/01/23
259 Allow RSYNC_PROXY to be http://user:pass@proxy.foo:3128/, and do
260 HTTP Basic Proxy-Authentication.
262 Multiple schemes are possible, up to and including the insanity that
263 is NTLM, but Basic probably covers most cases.
270 Add --with-socks, and then perhaps a command-line option to put them
271 on or off. This might be more reliable than LD_PRELOAD hacks.
278 rsync to a FAT partition on a Unix machine doesn't work very well at
279 the moment. I think we get errors about invalid filenames and
280 perhaps also trying to do atomic renames.
282 I guess the code to do this is currently #ifdef'd on Windows;
283 perhaps we ought to intelligently fall back to it on Unix too.
288 Allow forcing arbitrary permissions 2002/03/12
290 On 12 Mar 2002, Dave Dykstra <dwd@bell-labs.com> wrote:
291 > If we would add an option to do that functionality, I
292 > would vote for one that was more general which could mask
293 > off any set of permission bits and possibly add any set of
294 > bits. Perhaps a chmod-like syntax if it could be
295 > implemented simply.
297 I think that would be good too. For example, people uploading files
298 to a web server might like to say
300 rsync -avzP --chmod a+rX ./ sourcefrog.net:/home/www/sourcefrog/
302 Ideally the patch would implement as many of the gnu chmod semantics
303 as possible. I think the mode parser should be a separate function
304 that passes back something like (mask,set) description to the rest
305 of the program. For bonus points there would be a test case for the
308 Possibly also --chown
315 --diff david.e.sewell 2002/03/15
317 Allow people to specify the diff command. (Might want to use wdiff,
320 Just diff the temporary file with the destination file, and delete
321 the tmp file rather than moving it into place.
323 Interaction with --partial.
325 Security interactions with daemon mode?
330 Add daemon --no-detach and --no-fork options
332 Very useful for debugging. Also good when running under a
333 daemon-monitoring process that tries to restart the service when the
339 Create more granular verbosity jw 2003/05/15
341 Control output with the --report option.
343 The option takes as a single argument (no whitespace) a
344 comma delimited lists of keywords.
346 This would separate debugging from "logging" as well as
347 fine grained selection of statistical reporting and what
350 http://lists.samba.org/archive/rsync/2003-May/006059.html
354 DOCUMENTATION --------------------------------------------------------
361 Keep list of open issues and todos on the web site
366 Update web site from CVS
371 Perhaps redo manual as SGML
373 The man page is getting rather large, and there is more information
374 that ought to be added.
376 TexInfo source is probably a dying format.
378 Linuxdoc looks like the most likely contender. I know DocBook is
379 favoured by some people, but it's so bloody verbose, even with emacs
384 LOGGING --------------------------------------------------------------
386 Make dry run list all updates 2002/04/03
390 Mark Santcroos points out that -n fails to list files which have
391 only metadata changes, though it probably should.
393 There may be a Debian bug about this as well.
400 At exit, show how much memory was used for the file list, etc.
402 Also we do a wierd exponential-growth allocation in flist.c. I'm
403 not sure this makes sense with modern mallocs. At any rate it will
404 make us allocate a huge amount of memory for large file lists.
409 Improve error messages
411 If we hang or get SIGINT, then explain where we were up to. Perhaps
412 have a static buffer that contains the current function name, or
413 some kind of description of what we were trying to do. This is a
414 little easier on people than needing to run strace/truss.
416 "The dungeon collapses! You are killed." Rather than "unexpected
417 eof" give a message that is more detailed if possible and also more
420 If we get an error writing to a socket, then we should perhaps
421 continue trying to read to see if an error message comes across
422 explaining why the socket is closed. I'm not sure if this would
423 work, but it would certainly make our messages more helpful.
425 What happens if a directory is missing -x attributes. Do we lose
426 our load? (Debian #28416) Probably fixed now, but a test case would
434 Better statistics: Rasmus 2002/03/08
437 hey, how about an rsync option that just gives you the
438 summary without the list of files? And perhaps gives
439 more information like the number of new files, number
440 of changed, deleted, etc. ?
443 nice idea there is --stats but at the moment it's very
444 tridge-oriented rather than user-friendly it would be
445 nice to improve it that would also work well with
451 Perhaps flush stdout like syslog
453 Perhaps flush stdout after each filename, so that people trying to
454 monitor progress in a log file can do so more easily. See
455 http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=48108
460 Log deamon sessions that just list modules
462 At the connections that just get a list of modules are not logged,
468 Log child death on signal
470 If a child of the rsync daemon dies with a signal, we should notice
471 that when we reap it and log a message.
476 Keep stderr and stdout properly separated (Debian #23626)
481 Log errors with function that reports process of origin
483 Use a separate function for reporting errors; prefix it with
484 "rsync:" or "rsync(remote)", or perhaps even "rsync(local
490 verbose output David Stein 2001/12/20
492 Indicate whether files are new, updated, or deleted
494 At end of transfer, show how many files were or were not transferred
500 Add reason for transfer to file logging
502 Explain *why* every file is transferred or not (e.g. "local mtime
503 123123 newer than 1283198")
508 debugging of daemon 2002/04/08
510 Add an rsyncd.conf parameter to turn on debugging on the server.
517 Change to using gettext(). Probably need to ship this for platforms
520 Solicit translations.
522 Does anyone care? Before we bother modifying the code, we ought to
523 get the manual translated first, because that's possibly more useful
524 and at any rate demonstrates desire.
528 DEVELOPMENT --------------------------------------------------------
530 Handling duplicate names
532 We need to be careful of duplicate names getting into the file list.
533 See clean_flist(). This could happen if multiple arguments include
536 I think duplicates are only a problem if they're both flowing
537 through the pipeline at the same time. For example we might have
538 updated the first occurrence after reading the checksums for the
539 second. So possibly we just need to make sure that we don't have
540 both in the pipeline at the same time.
542 Possibly if we did one directory at a time that would be sufficient.
544 Alternatively we could pre-process the arguments to make sure no
545 duplicates will ever be inserted. There could be some bad cases
546 when we're collapsing symlinks.
548 We could have a hash table.
550 The root of the problem is that we do not want more than one file
551 list entry referring to the same file. At first glance there are
552 several ways this could happen: symlinks, hardlinks, and repeated
553 names on the command line.
555 If names are repeated on the command line, they may be present in
556 different forms, perhaps by traversing directory paths in different
557 ways, traversing paths including symlinks. Also we need to allow
558 for expansion of globs by rsync.
560 At the moment, clean_flist() requires having the entire file list in
561 memory. Duplicate names are detected just by a string comparison.
563 We don't need to worry about hard links causing duplicates because
564 files are never updated in place. Similarly for symlinks.
566 I think even if we're using a different symlink mode we don't need
569 Unless we're really clever this will introduce a protocol
570 incompatibility, so we need to be able to accept the old format as
576 Use generic zlib 2002/02/25
578 Perhaps don't use our own zlib.
582 - will automatically be up to date with bugfixes in zlib
584 - can leave it out for small rsync on e.g. recovery disks
586 - can use a shared library
588 - avoids people breaking rsync by trying to do this themselves and
591 Should we ship zlib for systems that don't have it, or require
592 people to install it separately?
594 Apparently this will make us incompatible with versions of rsync
595 that use the patched version of rsync. Probably the simplest way to
596 do this is to just disable gzip (with a warning) when talking to old
604 Rather than storing the file list in memory, store it in a TDB.
606 This *might* make memory usage lower while building the file list.
608 Hashtable lookup will mean files are not transmitted in order,
611 This would neatly eliminate one of the major post-fork shared data
619 Build rsync with SPLINT to try to find security holes. Add
620 annotations as necessary. Keep track of the number of warnings
621 found initially, and see how many of them are real bugs, or real
622 security bugs. Knowing the percentage of likely hits would be
623 really interesting for other projects.
630 jra recommends Valgrind:
632 http://devel-home.kde.org/~sewardj/
637 Create release script
643 Build tar file; upload
645 Send announcement to mailing list and c.o.l.a.
647 Make freshmeat announcement
654 Add machines to build farm
656 Cygwin (on different versions of Win32?)
658 HP-UX variants (via HP?)
666 PERFORMANCE ----------------------------------------------------------
668 File list structure in memory
670 Rather than one big array, perhaps have a tree in memory mirroring
673 This might make sorting much faster! (I'm not sure it's a big CPU
676 It might also reduce memory use in storing repeated directory names
677 -- again I'm not sure this is a problem.
682 Traverse just one directory at a time
684 Traverse just one directory at a time. Tridge says it's possible.
686 At the moment rsync reads the whole file list into memory at the
687 start, which makes us use a lot of memory and also not pipeline
688 network access as much as we could.
693 Allow skipping MD4 file_sum 2002/04/08
695 If we're doing a local transfer, or using -W, then perhaps don't
696 send the file checksum. If we're doing a local transfer, then
697 calculating MD4 checksums uses 90% of CPU and is unlikely to be
700 Indeed for transfers over zlib or ssh we can also rely on the
701 transport to have quite strong protection against corruption.
703 Perhaps we should have an option to disable this,
704 analogous to --whole-file, although it would default to
705 disabled. The file checksum takes up a definite space in
706 the protocol -- we can either set it to 0, or perhaps just
714 Perhaps borrow an assembler MD4 from someone?
716 Make sure we call MD4 with properly-sized blocks whenever possible
717 to avoid copying into the residue region?
721 TESTING --------------------------------------------------------------
725 Something that just keeps running rsync continuously over a data set
726 likely to generate problems.
731 Cross-test versions 2001/08/22
733 Part of the regression suite should be making sure that we
734 don't break backwards compatibility: old clients vs new
735 servers and so on. Ideally we would test both up and down
736 from the current release to all old versions.
738 Run current rsync versions against significant past releases.
740 We might need to omit broken old versions, or versions in which
741 particular functionality is broken
743 It might be sufficient to test downloads from well-known public
744 rsync servers running different versions of rsync. This will give
745 some testing and also be the most common case for having different
746 versions and not being able to upgrade.
748 The new --protocol option may help in this.
753 Test on kernel source
755 Download all versions of kernel; unpack, sync between them. Also
756 sync between uncompressed tarballs. Compare directories after
759 Use local mode; ssh; daemon; --whole-file and --no-whole-file.
761 Use awk to pull out the 'speedup' number for each transfer. Make
769 Sparse and non-sparse
774 Create mutator program for testing
776 Insert bytes, delete bytes, swap blocks, ...
781 Create configure option to enable dangerous tests
786 If tests are skipped, say why.
791 Test daemon feature to disallow particular options.
796 Create pipe program for testing
798 Create pipe program that makes slow/jerky connections for
799 testing Versions of read() and write() that corrupt the
800 stream, or abruptly fail
805 Create test makefile target for some tests
807 Separate makefile target to run rough tests -- or perhaps
808 just run them every time?
813 Test "refuse options" works
815 What about for --recursive?
817 If you specify an unrecognized option here, you should get an error.
819 We need a test case for this...
821 Was this broken when we changed to popt?
825 RELATED PROJECTS -----------------------------------------------------
829 Write a small emulation of interactive ftp as a Pythonn program
830 that calls rsync. Commands such as "cd", "ls", "ls *.c" etc map
831 fairly directly into rsync commands: it just needs to remember the
832 current host, directory and so on. We can probably even do
833 completion of remote filenames.
838 http://rsync.samba.org/rsync-and-debian/
846 Exhaustive, tortuous testing
853 rsyncsplit as alternative to real integration with gzip?
858 reverse rsync over HTTP Range
860 Goswin Brederlow suggested this on Debian; I think tridge and I
861 talked about it previous in relation to rproxy.