rsync/rsync.git
22 years agoGet rid of global_opts struct as suggested by Dave -- too many
Martin Pool [Tue, 18 Dec 2001 05:54:57 +0000 (05:54 +0000)]
Get rid of global_opts struct as suggested by Dave -- too many
problems with initialization.

Change the algorithm from trying to open an inbound socket with
getaddrinfo: keep trying suggested addresses until we find one on
which we can both get a socket and bind.  Not convinced this is the
best, but it's probably better.

22 years agoNote rsyncd-over-ssh and documentation TODOs.
Martin Pool [Tue, 18 Dec 2001 01:33:56 +0000 (01:33 +0000)]
Note rsyncd-over-ssh and documentation TODOs.

22 years agoFix from Jeff Garzik for inet_ntop prototype errors on some Linux
Martin Pool [Tue, 18 Dec 2001 01:32:27 +0000 (01:32 +0000)]
Fix from Jeff Garzik for inet_ntop prototype errors on some Linux
distributions: we were failing to define HAVE_INET_NTOP, so our
prototype in rsync.h came through.

Also rerun autoheader, and have comment for HAVE_SOCKADDR_STORAGE.

22 years agoUpdate README so that Paulus doesn't get bug reports anymore :-)
Martin Pool [Tue, 18 Dec 2001 01:07:20 +0000 (01:07 +0000)]
Update README so that Paulus doesn't get bug reports anymore :-)

I think this document needs to be reworked to better explain how to
use rsync, but not right now.

22 years agoWhen INET6 is not defined, meaning that IPv6 is not supported, need to
David Dykstra [Fri, 14 Dec 2001 18:25:51 +0000 (18:25 +0000)]
When INET6 is not defined, meaning that IPv6 is not supported, need to
initalize the global_opts.af_hint to AF_INET or systems such as Linux that
have a native getaddrinfo() because they support IPv6 will attempt to
create IPv6 sockets.  This brings up a problem with the new global_opts
structure; in order to initialize them to a value other than 0, we need to
explicitly initialize them all in an order that matches the order in
rsync.h.  I think that's more inconvenient & error prone than keeping
global variables.

22 years agoopen_socket_in was attempting to try all the protocols returned from
David Dykstra [Fri, 14 Dec 2001 18:00:54 +0000 (18:00 +0000)]
open_socket_in was attempting to try all the protocols returned from
getaddrinfo(), but only if a corresponding call to socket() returned one of
three *NOSUPPORT errno codes.  A Redhat 6.2 system was observed returning
EINVAL instead so it never went on to try IPv4.  This update adds EINVAL to
the list.  Question: why not always continue through the list regardless of
what errno is?

22 years agoVerbose.
Martin Pool [Fri, 14 Dec 2001 05:55:21 +0000 (05:55 +0000)]
Verbose.

22 years agoOnly show test output if it failed.
Martin Pool [Fri, 14 Dec 2001 05:54:24 +0000 (05:54 +0000)]
Only show test output if it failed.

22 years agoBe less verbose.
Martin Pool [Fri, 14 Dec 2001 05:52:25 +0000 (05:52 +0000)]
Be less verbose.

22 years agoFix quoting.
Martin Pool [Fri, 14 Dec 2001 05:44:04 +0000 (05:44 +0000)]
Fix quoting.

22 years agoPerhaps Solaris sh wants us to explicitly exit 0, rather than just
Martin Pool [Fri, 14 Dec 2001 05:27:42 +0000 (05:27 +0000)]
Perhaps Solaris sh wants us to explicitly exit 0, rather than just
using the last return code?

22 years agoMore error messages.
Martin Pool [Fri, 14 Dec 2001 05:19:15 +0000 (05:19 +0000)]
More error messages.

22 years agoTypo.
Martin Pool [Fri, 14 Dec 2001 05:17:52 +0000 (05:17 +0000)]
Typo.

22 years agoMake scratch directory properly.
Martin Pool [Fri, 14 Dec 2001 05:01:48 +0000 (05:01 +0000)]
Make scratch directory properly.

22 years agoDoc.
Martin Pool [Fri, 14 Dec 2001 02:18:31 +0000 (02:18 +0000)]
Doc.

22 years agoClearer error messages.
Martin Pool [Fri, 14 Dec 2001 02:16:20 +0000 (02:16 +0000)]
Clearer error messages.

22 years agoClearer error messages.
Martin Pool [Fri, 14 Dec 2001 02:14:58 +0000 (02:14 +0000)]
Clearer error messages.

22 years agoAdd a test that when none of -l, -L, -a are specified symlinks are not
Martin Pool [Fri, 14 Dec 2001 02:01:46 +0000 (02:01 +0000)]
Add a test that when none of -l, -L, -a are specified symlinks are not
copied at all.

22 years agoClean scratch directory between each test run.
Martin Pool [Fri, 14 Dec 2001 01:52:43 +0000 (01:52 +0000)]
Clean scratch directory between each test run.

22 years agoReadlink(2) does not nul-terminate the output buffer, so we were
Martin Pool [Fri, 14 Dec 2001 01:48:48 +0000 (01:48 +0000)]
Readlink(2) does not nul-terminate the output buffer, so we were
getting corrupt output when listing more than one symlink.

22 years agoFix contact details.
Martin Pool [Fri, 14 Dec 2001 01:09:23 +0000 (01:09 +0000)]
Fix contact details.

22 years agoDon't define DEBUG with --enable-debug, because that makes zlib emit strange messages
Martin Pool [Sun, 9 Dec 2001 21:48:13 +0000 (21:48 +0000)]
Don't define DEBUG with --enable-debug, because that makes zlib emit strange messages

22 years agoIRIX cc cares that the operands to the ternary have the same type.
Martin Pool [Wed, 5 Dec 2001 13:48:41 +0000 (13:48 +0000)]
IRIX cc cares that the operands to the ternary have the same type.

22 years agoBe more strict about 'enum logcode' rather than int. IRIX compiler
Martin Pool [Wed, 5 Dec 2001 13:45:51 +0000 (13:45 +0000)]
Be more strict about 'enum logcode' rather than int.  IRIX compiler
picked this up -- quite neat.

22 years agoSome platforms don't have sa_family_t.
Martin Pool [Wed, 5 Dec 2001 13:44:37 +0000 (13:44 +0000)]
Some platforms don't have sa_family_t.

22 years agoBump version
Martin Pool [Wed, 5 Dec 2001 13:41:38 +0000 (13:41 +0000)]
Bump version

22 years agoRFC2553 just says that sockaddr_storage has to have initial fields
Martin Pool [Wed, 5 Dec 2001 13:25:29 +0000 (13:25 +0000)]
RFC2553 just says that sockaddr_storage has to have initial fields
isomorphic to sa_family etc from a struct sockaddr, not what they're
called.  On some platforms they seem not to be called ss_family.
Rather than guess, we will try casting to a sockaddr and looking
through that -- I think this is what the RFC intends.

22 years agoSome platforms may have getaddrinfo() but not sockaddr_storage.
Martin Pool [Wed, 5 Dec 2001 13:19:16 +0000 (13:19 +0000)]
Some platforms may have getaddrinfo() but not sockaddr_storage.

22 years agoDocument --no-detach.
Martin Pool [Wed, 5 Dec 2001 13:10:24 +0000 (13:10 +0000)]
Document --no-detach.

22 years agoFix help message for --address option
Martin Pool [Wed, 5 Dec 2001 13:06:26 +0000 (13:06 +0000)]
Fix help message for --address option

22 years agoAdd --no-detach option for W32, daemontools, etc.
Martin Pool [Wed, 5 Dec 2001 13:03:16 +0000 (13:03 +0000)]
Add --no-detach option for W32, daemontools, etc.

22 years agoDoc
Martin Pool [Wed, 5 Dec 2001 13:02:11 +0000 (13:02 +0000)]
Doc

22 years agoDoc.
Martin Pool [Wed, 5 Dec 2001 12:56:06 +0000 (12:56 +0000)]
Doc.

22 years agoChange back to using sockaddr_storage rather than sockaddr_in. If
Martin Pool [Wed, 5 Dec 2001 12:48:46 +0000 (12:48 +0000)]
Change back to using sockaddr_storage rather than sockaddr_in.  If
sockaddr_storage is not defined, then supply our own definition that
will hopefully satisfy RFC2553 but also compile on all supported
platforms.

Thankyou to YOSHIFUJI Hideaki, SUMIKAWA Munechika and Jun-ichiro
"itojun" Hagino.

22 years agoSuggestions from KAME IPv6 newsletter.
Martin Pool [Wed, 5 Dec 2001 12:25:36 +0000 (12:25 +0000)]
Suggestions from KAME IPv6 newsletter.

22 years agoNote from tpot that sighup should have standard behaviour of reinitializing.
Martin Pool [Wed, 5 Dec 2001 00:58:20 +0000 (00:58 +0000)]
Note from tpot that sighup should have standard behaviour of reinitializing.

22 years agoDon't print out the directory name twice in verbose mode; it was being
David Dykstra [Mon, 3 Dec 2001 18:37:33 +0000 (18:37 +0000)]
Don't print out the directory name twice in verbose mode; it was being
printed a second time when the modification time of the directory was
being set, and that time around recv_files() calls recv_generator() with
an f_out of -1 so check that before printing the directory name.

22 years agoNote about crash from Ayamura KIKUCHI
Martin Pool [Sun, 2 Dec 2001 22:47:30 +0000 (22:47 +0000)]
Note about crash from Ayamura KIKUCHI

22 years agobatch reindent
Martin Pool [Sun, 2 Dec 2001 22:28:50 +0000 (22:28 +0000)]
batch reindent

22 years agoBump version
Martin Pool [Sun, 2 Dec 2001 22:26:09 +0000 (22:26 +0000)]
Bump version

22 years agoFix FP usage.
Martin Pool [Sun, 2 Dec 2001 14:12:03 +0000 (14:12 +0000)]
Fix FP usage.

22 years agoOops, units bug.
Martin Pool [Sun, 2 Dec 2001 13:58:03 +0000 (13:58 +0000)]
Oops, units bug.

22 years agoShow time-remaining as hh:mm:ss.
Martin Pool [Sun, 2 Dec 2001 13:45:38 +0000 (13:45 +0000)]
Show time-remaining as hh:mm:ss.

22 years agoUpdate notes
Martin Pool [Sun, 2 Dec 2001 13:22:55 +0000 (13:22 +0000)]
Update notes

22 years agoImproved estimation algorithm for time-to-complete.
Martin Pool [Sun, 2 Dec 2001 13:07:07 +0000 (13:07 +0000)]
Improved estimation algorithm for time-to-complete.

22 years agoTypo fix by Tom Schmidt
Martin Pool [Sun, 2 Dec 2001 13:02:17 +0000 (13:02 +0000)]
Typo fix by Tom Schmidt

22 years agoconst-cast required for silly UNICOS headers
Martin Pool [Sun, 2 Dec 2001 12:37:48 +0000 (12:37 +0000)]
const-cast required for silly UNICOS headers

22 years agoAlso estimate time remaining.
Martin Pool [Sun, 2 Dec 2001 08:56:25 +0000 (08:56 +0000)]
Also estimate time remaining.

22 years agoProgress indicator now shows estimated rate of transfer (e.g. kB/s).
Martin Pool [Sun, 2 Dec 2001 08:38:51 +0000 (08:38 +0000)]
Progress indicator now shows estimated rate of transfer (e.g. kB/s).

Based on a patch from Rik Faith, but modified to make sure we do only
one rprintf call, and that we never end up with two copies of the line
printed out.

22 years agoRevert change from 1.39, because it causes a crash because of
Martin Pool [Sun, 2 Dec 2001 08:16:15 +0000 (08:16 +0000)]
Revert change from 1.39, because it causes a crash because of
attempting to free a static string.  (Thankyou to Paul Mackerras.)
There's still a small leak here.

22 years agoFix test suite breakage in calling tls.
Martin Pool [Sun, 2 Dec 2001 07:22:54 +0000 (07:22 +0000)]
Fix test suite breakage in calling tls.

Clean up test directory on completion.

22 years agoBump version
Martin Pool [Sun, 2 Dec 2001 07:21:55 +0000 (07:21 +0000)]
Bump version

22 years agoMove old news from release 2.5.0.
Martin Pool [Sun, 2 Dec 2001 07:17:50 +0000 (07:17 +0000)]
Move old news from release 2.5.0.

22 years agoMore notes from email.
Martin Pool [Sun, 2 Dec 2001 07:07:43 +0000 (07:07 +0000)]
More notes from email.

22 years agoTypo fix from Matt Kraai <kraai@debian.org>
Martin Pool [Sun, 2 Dec 2001 06:17:33 +0000 (06:17 +0000)]
Typo fix from Matt Kraai <kraai@debian.org>

22 years agoAdd note about device major/minor numbers, and about ACLs
Martin Pool [Sun, 2 Dec 2001 05:12:39 +0000 (05:12 +0000)]
Add note about device major/minor numbers, and about ACLs

22 years agoAdd "$(OBJS): config.h" so everything will rebuild if config.h changes.
David Dykstra [Fri, 30 Nov 2001 22:21:20 +0000 (22:21 +0000)]
Add "$(OBJS): config.h" so everything will rebuild if config.h changes.

22 years agoAdd message to ignore warnings about mktemp
Martin Pool [Fri, 30 Nov 2001 22:06:29 +0000 (22:06 +0000)]
Add message to ignore warnings about mktemp

22 years agoModified file stored as well
rsync-bugs [Fri, 30 Nov 2001 00:31:06 +0000 (00:31 +0000)]
Modified file stored as well

22 years agopreparing for release of 2.5.0
rsync-bugs [Fri, 30 Nov 2001 00:29:46 +0000 (00:29 +0000)]
preparing for release of 2.5.0

22 years agoSpec file needs to be generated from a template to include the version
Martin Pool [Fri, 30 Nov 2001 00:29:20 +0000 (00:29 +0000)]
Spec file needs to be generated from a template to include the version
number.

22 years agopreparing for release of 2.5.0
rsync-bugs [Fri, 30 Nov 2001 00:23:11 +0000 (00:23 +0000)]
preparing for release of 2.5.0

22 years agoLast bug fix for 2.5.0? Make the documentation for -a point out that
Martin Pool [Fri, 30 Nov 2001 00:17:53 +0000 (00:17 +0000)]
Last bug fix for 2.5.0?  Make the documentation for -a point out that
it does not in fact propagate hardlinsk.

22 years agoNote error cases that we ought to improve.
Martin Pool [Fri, 30 Nov 2001 00:16:14 +0000 (00:16 +0000)]
Note error cases that we ought to improve.

22 years agoNote other supported platforms
Martin Pool [Fri, 30 Nov 2001 00:15:32 +0000 (00:15 +0000)]
Note other supported platforms

22 years agoAdd notes on things to do mentioned on the list in the last few months.
Martin Pool [Thu, 29 Nov 2001 01:31:31 +0000 (01:31 +0000)]
Add notes on things to do mentioned on the list in the last few months.

22 years agoIPv6 support is now merged.
Martin Pool [Thu, 29 Nov 2001 00:23:29 +0000 (00:23 +0000)]
IPv6 support is now merged.

22 years agoJust include rsync.h rather than all the individual headers.
Martin Pool [Thu, 29 Nov 2001 00:15:20 +0000 (00:15 +0000)]
Just include h rather than all the individual headers.
(Suggestion from Dave Dykstra.)

22 years agoNo need to test for memcmp, because even broken ones are good enough
Martin Pool [Thu, 29 Nov 2001 00:08:36 +0000 (00:08 +0000)]
No need to test for memcmp, because even broken ones are good enough
for our purposes.  (Patch from Dave Dykstra.)

22 years agoShow version when configuring.
Martin Pool [Thu, 29 Nov 2001 00:04:48 +0000 (00:04 +0000)]
Show version when configuring.

If we don't seem to have an ANSI compiler, then omit a warning as soon
as that is discovered, because it is likely to break later configure
tests.  This doesn't seem to catch the particular HP-UX compiler I was
after, which is non-ANSI but only emits a warning on this configure
test.  Nevertheless probably better to have it in.

22 years agoAdd list of platforms that build.
Martin Pool [Wed, 28 Nov 2001 07:49:32 +0000 (07:49 +0000)]
Add list of platforms that build.

22 years agoNote half-baked Darwin IPv6 support.
Martin Pool [Wed, 28 Nov 2001 07:12:09 +0000 (07:12 +0000)]
Note half-baked Darwin IPv6 support.

22 years agoPut back the --disable-ipv6 option. This should only be needed if
Martin Pool [Wed, 28 Nov 2001 06:52:04 +0000 (06:52 +0000)]
Put back the --disable-ipv6 option.  This should only be needed if
your platform seems to support ipv6, but actually it breaks.  This
seems to be the case for "powerpc-apple-darwin1.4".

22 years agoCall this 2.4.7pre4
Martin Pool [Wed, 28 Nov 2001 04:32:33 +0000 (04:32 +0000)]
Call this 2.4.7pre4

22 years agoNote on HP-UX's bundled so-called C compiler.
Martin Pool [Wed, 28 Nov 2001 04:10:39 +0000 (04:10 +0000)]
Note on HP-UX's bundled so-called C compiler.

22 years agoHP's cpp apparently can't handle whitespace before #include
Martin Pool [Wed, 28 Nov 2001 03:12:20 +0000 (03:12 +0000)]
HP's cpp apparently can't handle whitespace before #include

22 years agoHP's cpp chokes on preprocessor directives that have whitespace before
Martin Pool [Wed, 28 Nov 2001 02:55:20 +0000 (02:55 +0000)]
HP's cpp chokes on preprocessor directives that have whitespace before
the '#'.  Off to the Implant Office with you!

22 years agoFix inet_ntop/pton names
Martin Pool [Wed, 28 Nov 2001 01:29:41 +0000 (01:29 +0000)]
Fix inet_ntop/pton names

22 years agoProtect AF_INET6 references with #ifdef INET6
Martin Pool [Wed, 28 Nov 2001 00:36:32 +0000 (00:36 +0000)]
Protect AF_INET6 references with #ifdef INET6

22 years agoFix sh
Martin Pool [Tue, 27 Nov 2001 07:59:34 +0000 (07:59 +0000)]
Fix sh

22 years agoShow symlink targets
Martin Pool [Tue, 27 Nov 2001 07:56:33 +0000 (07:56 +0000)]
Show symlink targets

22 years agoBe a bit more verbose
Martin Pool [Tue, 27 Nov 2001 07:54:48 +0000 (07:54 +0000)]
Be a bit more verbose

22 years agomtime and ownership of symlinks can be random, so don't print them.
Martin Pool [Tue, 27 Nov 2001 07:54:03 +0000 (07:54 +0000)]
mtime and ownership of symlinks can be random, so don't print them.

22 years agoBe a bit more verbose
Martin Pool [Tue, 27 Nov 2001 07:32:42 +0000 (07:32 +0000)]
Be a bit more verbose

22 years agoTry to fix headers for UNICOS
Martin Pool [Tue, 27 Nov 2001 07:23:58 +0000 (07:23 +0000)]
Try to fix headers for UNICOS

22 years agoUnbreak recursive ls test
Martin Pool [Tue, 27 Nov 2001 07:19:37 +0000 (07:19 +0000)]
Unbreak recursive ls test

22 years agoOops, fix bash syntax
Martin Pool [Tue, 27 Nov 2001 07:09:47 +0000 (07:09 +0000)]
Oops, fix bash syntax

22 years agoThe size of anything but a regular file is probably not worth thinking
Martin Pool [Tue, 27 Nov 2001 07:07:36 +0000 (07:07 +0000)]
The size of anything but a regular file is probably not worth thinking
about.

22 years agoWhen producing a ls-style permissions string, also handle
Martin Pool [Tue, 27 Nov 2001 07:05:33 +0000 (07:05 +0000)]
When producing a ls-style permissions string, also handle
sticky/setuid/setgid bits the same way as GNU ls.

22 years agoUse tls rather than the OS's ls(1) so that we have more chance of
Martin Pool [Tue, 27 Nov 2001 06:51:51 +0000 (06:51 +0000)]
Use tls rather than the OS's ls(1) so that we have more chance of
reproducible results.

22 years agoAlso list permissions, ownership, size, and mtime.
Martin Pool [Tue, 27 Nov 2001 06:45:47 +0000 (06:45 +0000)]
Also list permissions, ownership, size, and mtime.

22 years agoSplit code to generate "rwx-----" strings into lib/permstring.c so it
Martin Pool [Tue, 27 Nov 2001 06:43:17 +0000 (06:43 +0000)]
Split code to generate "rwx-----" strings into lib/permstring.c so it
can be reused in tls.

22 years agoBuild getaddr/nameinfo into lib/.
Martin Pool [Tue, 27 Nov 2001 06:41:38 +0000 (06:41 +0000)]
Build getaddr/nameinfo into lib/.

22 years agomake clean can rm config.cache but not config.h, or it wil jam.
Martin Pool [Tue, 27 Nov 2001 06:39:02 +0000 (06:39 +0000)]
make clean can rm config.cache but not config.h, or it wil jam.

Build getaddr/nameinfo into lib/.

Split code to generate "rwx-----" strings into lib/permstring.c so it
can be reused in tls.

22 years agoTry to fix LIBOBJ detection of ntop/pton.
Martin Pool [Tue, 27 Nov 2001 06:17:33 +0000 (06:17 +0000)]
Try to fix LIBOBJ detection of ntop/pton.

22 years agoFix missing parameter in log call.
Martin Pool [Tue, 27 Nov 2001 06:01:05 +0000 (06:01 +0000)]
Fix missing parameter in log call.

22 years agoMake clean should also remove the autoconf cache etc.
Martin Pool [Tue, 27 Nov 2001 05:35:14 +0000 (05:35 +0000)]
Make clean should also remove the autoconf cache etc.

22 years agoAdd Paul Vixie's implementation of inet_ntop and inet_pton for
Martin Pool [Tue, 27 Nov 2001 05:22:23 +0000 (05:22 +0000)]
Add Paul Vixie's implementation of inet_ntop and inet_pton for
platforms that don't have them.

22 years agoLook in -lresolv for inet_ntop
Martin Pool [Tue, 27 Nov 2001 04:53:08 +0000 (04:53 +0000)]
Look in -lresolv for inet_ntop

22 years agoCheck for inet_ntop and inet_pton, which may be missing.
Martin Pool [Tue, 27 Nov 2001 03:49:53 +0000 (03:49 +0000)]
Check for inet_ntop and inet_pton, which may be missing.