rsync.git
14 years ago- Changed log_open() into two functions, logfile_open() and
Wayne Davison [Thu, 14 Apr 2005 16:08:10 +0000 (16:08 +0000)]
- Changed log_open() into two functions, logfile_open() and
  logfile_reopen().
- Changed log_close() into logfile_close().
- Improved the logic in logit().

14 years agoMention that less data is sent over the wire when --only-write-batch
Wayne Davison [Thu, 14 Apr 2005 01:47:47 +0000 (01:47 +0000)]
Mention that less data is sent over the wire when --only-write-batch
is used and we're pushing files to a remote system.

14 years agoGot rid of ITEM_DUMMY_BIT.
Wayne Davison [Thu, 14 Apr 2005 01:46:08 +0000 (01:46 +0000)]
Got rid of ITEM_DUMMY_BIT.

14 years ago- Got rid of the iflag kluge for 2.6.4pre3.
Wayne Davison [Thu, 14 Apr 2005 01:45:47 +0000 (01:45 +0000)]
- Got rid of the iflag kluge for 2.6.4pre3.
- When we're handling --only-write-batch on the client side, we now
  send all index/header info to the receiver (not just to the batch
  file, like we do with the xfer data) so that the connection cannot
  timeout.

14 years agoSupport the reception of a file-transfer header without xfer data
Wayne Davison [Thu, 14 Apr 2005 01:42:13 +0000 (01:42 +0000)]
Support the reception of a file-transfer header without xfer data
when we're on the server side and --only-write-batch was specified.

14 years agoMention how the daemon handles a failure to open a user-specified
Wayne Davison [Tue, 12 Apr 2005 23:09:35 +0000 (23:09 +0000)]
Mention how the daemon handles a failure to open a user-specified
"log file".

14 years agoMention how we now handle a log-file that the daemon can't open.
Wayne Davison [Tue, 12 Apr 2005 23:06:45 +0000 (23:06 +0000)]
Mention how we now handle a log-file that the daemon can't open.

14 years agoGot rid of RERR_LOG_FAILURE.
Wayne Davison [Tue, 12 Apr 2005 23:04:10 +0000 (23:04 +0000)]
Got rid of RERR_LOG_FAILURE.

14 years agoIf we can't open the daemon's log file, fall-back to using
Wayne Davison [Tue, 12 Apr 2005 23:03:49 +0000 (23:03 +0000)]
If we can't open the daemon's log file, fall-back to using
syslog (rather than trying to output an error on stderr,
which would usually be lost, and then exiting).

14 years agoA couple improvements for the --only-write-batch section.
Wayne Davison [Tue, 12 Apr 2005 03:55:56 +0000 (03:55 +0000)]
A couple improvements for the --only-write-batch section.

14 years agoMention the latest bug-fix.
Wayne Davison [Sun, 10 Apr 2005 20:08:00 +0000 (20:08 +0000)]
Mention the latest bug-fix.

14 years agoPaul's patch to improve the "not creating new ..." message
Wayne Davison [Sun, 10 Apr 2005 19:44:16 +0000 (19:44 +0000)]
Paul's patch to improve the "not creating new ..." message
for directories.

14 years agoMade the code a little cleaner by having gen_challenge() return
Wayne Davison [Sun, 10 Apr 2005 17:09:10 +0000 (17:09 +0000)]
Made the code a little cleaner by having gen_challenge() return
the challenge string base64-encoded (instead of forcing the caller
to handle this).  Also improved a couple comments.

14 years agoMention fix for SUPPORT{,_HARD}_LINKS defines.
Wayne Davison [Sun, 10 Apr 2005 07:17:47 +0000 (07:17 +0000)]
Mention fix for SUPPORT{,_HARD}_LINKS defines.

14 years ago- Don't define HAVE_READLINK unless HAVE_READLINK is defined.
Wayne Davison [Sun, 10 Apr 2005 06:24:14 +0000 (06:24 +0000)]
- Don't define HAVE_READLINK unless HAVE_READLINK is defined.
- Don't define SUPPORT_HARD_LINKS unless HAVE_LINK is defined.

14 years agoMention the new --only-write-batch=FILE option.
Wayne Davison [Sat, 9 Apr 2005 18:59:59 +0000 (18:59 +0000)]
Mention the new --only-write-batch=FILE option.

14 years ago- Use the new do_xfers variable in place of some dry_run checks
Wayne Davison [Sat, 9 Apr 2005 18:59:57 +0000 (18:59 +0000)]
- Use the new do_xfers variable in place of some dry_run checks
  (with appropriate negation).
- If write_batch is < 0, we write out the file-transfer data to
  just the batch file, not the socket.

14 years ago- Use the new do_xfers variable in place of some dry_run checks
Wayne Davison [Sat, 9 Apr 2005 18:59:55 +0000 (18:59 +0000)]
- Use the new do_xfers variable in place of some dry_run checks
  (with appropriate negation).
- If we get a transfer when write_batch is < 0, discard it.

14 years agoUse the new do_xfers variable in place of some dry_run checks
Wayne Davison [Sat, 9 Apr 2005 18:59:52 +0000 (18:59 +0000)]
Use the new do_xfers variable in place of some dry_run checks
(with appropriate negation).

14 years ago- Only do read-/write-batch processing if we're not the server (needed
Wayne Davison [Sat, 9 Apr 2005 18:59:49 +0000 (18:59 +0000)]
- Only do read-/write-batch processing if we're not the server (needed
  now that write_batch is no longer forced to 0 for the server).
- If write_batch is < 0, force "dry_run = 1" (but only after we've
  finished any appropriate write-batch file creation).

14 years ago- Added do_xfers variable.
Wayne Davison [Sat, 9 Apr 2005 18:59:47 +0000 (18:59 +0000)]
- Added do_xfers variable.
- Handle the new --only-write-batch=FILE option.

14 years agoAlso change --only-write-batch to --read-batch (in addition to
Wayne Davison [Sat, 9 Apr 2005 18:59:44 +0000 (18:59 +0000)]
Also change --only-write-batch to --read-batch (in addition to
changing --write-batch).

14 years agoWe no longer force write_batch to 0 for the server.
Wayne Davison [Sat, 9 Apr 2005 18:59:42 +0000 (18:59 +0000)]
We no longer force write_batch to 0 for the server.

14 years agoDocument the new --only-write-batch=FILE option.
Wayne Davison [Sat, 9 Apr 2005 18:59:40 +0000 (18:59 +0000)]
Document the new --only-write-batch=FILE option.

14 years agoMention the latest changes.
Wayne Davison [Sat, 9 Apr 2005 18:15:22 +0000 (18:15 +0000)]
Mention the latest changes.

14 years ago- Call auth_server() with its new "host" arg.
Wayne Davison [Sat, 9 Apr 2005 18:11:25 +0000 (18:11 +0000)]
- Call auth_server() with its new "host" arg.
- Don't log an auth-failed error -- auth_server() now handles that.

14 years ago- Use the MD4_SUM_LENGTH define in place of some hard-wired values.
Wayne Davison [Sat, 9 Apr 2005 18:11:23 +0000 (18:11 +0000)]
- Use the MD4_SUM_LENGTH define in place of some hard-wired values.
- Pass the hostname in to auth_server().
- Generate a unique error for each failure type in auth_server() so
  that the log-file contains why the authorization failed.
- Don't use sscanf() to parse the client's auth-challenge response.

14 years agoThe "@ERROR" handler in start_inband_exchange() was not returning
Wayne Davison [Sat, 9 Apr 2005 16:49:51 +0000 (16:49 +0000)]
The "@ERROR" handler in start_inband_exchange() was not returning
the right value.

14 years agoFixed one typo and improved another sentence.
Wayne Davison [Fri, 8 Apr 2005 01:40:57 +0000 (01:40 +0000)]
Fixed one typo and improved another sentence.

14 years agoMake sure that "- !" or "+ !" aren't interpreted as a list-clearing
Wayne Davison [Thu, 7 Apr 2005 18:06:06 +0000 (18:06 +0000)]
Make sure that "- !" or "+ !" aren't interpreted as a list-clearing
token.

14 years agoCombine one of the backup tests with --delete-after.
Wayne Davison [Thu, 7 Apr 2005 09:09:17 +0000 (09:09 +0000)]
Combine one of the backup tests with --delete-after.

14 years agoDon't call maybe_flush_socket() quite so often.
Wayne Davison [Thu, 7 Apr 2005 08:32:32 +0000 (08:32 +0000)]
Don't call maybe_flush_socket() quite so often.

14 years agoMention the latest changes.
Wayne Davison [Thu, 7 Apr 2005 08:28:06 +0000 (08:28 +0000)]
Mention the latest changes.

14 years agoFix a bug with --delete-after combined with --backup.
Wayne Davison [Thu, 7 Apr 2005 08:13:44 +0000 (08:13 +0000)]
Fix a bug with --delete-after combined with --backup.

14 years agoUse the new HAVE_LSEEK64 define.
Wayne Davison [Wed, 6 Apr 2005 02:08:21 +0000 (02:08 +0000)]
Use the new HAVE_LSEEK64 define.

14 years agoAdded skipping of mkstemp() on HP-UX.
Wayne Davison [Wed, 6 Apr 2005 02:07:21 +0000 (02:07 +0000)]
Added skipping of mkstemp() on HP-UX.

14 years agoPeriodically call maybe_flush_socket().
Wayne Davison [Tue, 5 Apr 2005 20:08:51 +0000 (20:08 +0000)]
Periodically call maybe_flush_socket().

14 years ago- Added maybe_flush_socket() for use by the generator.
Wayne Davison [Tue, 5 Apr 2005 20:07:42 +0000 (20:07 +0000)]
- Added maybe_flush_socket() for use by the generator.
- Always maintain the last_io value for the generator.

14 years agoWhen doing a delete pass with do_progress output, get rid of the
Wayne Davison [Tue, 5 Apr 2005 19:51:13 +0000 (19:51 +0000)]
When doing a delete pass with do_progress output, get rid of the
temporary output (it used to always be covered up by output in
older versions, but it might be followed by a newline in newer
versions).

14 years agoDocument the use of --address in client mode.
Wayne Davison [Tue, 5 Apr 2005 06:00:17 +0000 (06:00 +0000)]
Document the use of --address in client mode.

14 years agoAllows --address to be used in client mode.
Wayne Davison [Tue, 5 Apr 2005 05:59:49 +0000 (05:59 +0000)]
Allows --address to be used in client mode.

14 years agoChanged the bind_address local variables to be named bind_addr
Wayne Davison [Tue, 5 Apr 2005 05:52:49 +0000 (05:52 +0000)]
Changed the bind_address local variables to be named bind_addr
in order to avoid confusion with the bind_address global.

14 years agoMake sure that the line buffer in readfd_unbuffered() is large
Wayne Davison [Mon, 4 Apr 2005 17:27:56 +0000 (17:27 +0000)]
Make sure that the line buffer in readfd_unbuffered() is large
enough to handle long-filename messages on a system that has a
really short MAXPATHLEN value.  Also, make it large enough to
be able to comment on a MAXPATHLEN filename.

14 years agoIn writefd_unbuffered(), make sure that we don't look at the r_fds
Wayne Davison [Mon, 4 Apr 2005 00:48:39 +0000 (00:48 +0000)]
In writefd_unbuffered(), make sure that we don't look at the r_fds
variable when we're ignoring msg_fd_in.

14 years agoMention fix for --compare-dest.
Wayne Davison [Fri, 1 Apr 2005 18:18:17 +0000 (18:18 +0000)]
Mention fix for --compare-dest.

14 years agoMake a local-copy caused by a not-quite-up-to-date --compare-dest
Wayne Davison [Fri, 1 Apr 2005 18:12:22 +0000 (18:12 +0000)]
Make a local-copy caused by a not-quite-up-to-date --compare-dest
or --link-dest file be output as a 'c' (local change) when itemizing
or as transfered file when not itemizing.

14 years ago- Conditionally include <locale.h>.
Wayne Davison [Fri, 1 Apr 2005 17:25:37 +0000 (17:25 +0000)]
- Conditionally include <locale.h>.
- Conditionally call setlocale(LC_CTYPE, "").

14 years ago- Changed version to 2.6.5cvs.
Wayne Davison [Fri, 1 Apr 2005 17:25:34 +0000 (17:25 +0000)]
- Changed version to 2.6.5cvs.
- Look for setlocale(), locale.h, and honor --disable-locale.

14 years agoStarting the 2.6.5 news.
Wayne Davison [Fri, 1 Apr 2005 17:25:28 +0000 (17:25 +0000)]
Starting the 2.6.5 news.

14 years agoMoved the 2.6.4 news here.
Wayne Davison [Fri, 1 Apr 2005 17:25:26 +0000 (17:25 +0000)]
Moved the 2.6.4 news here.

14 years agoChanged safe_fname() to output \### (octal) to escape non-printable
Wayne Davison [Fri, 1 Apr 2005 16:48:54 +0000 (16:48 +0000)]
Changed safe_fname() to output \### (octal) to escape non-printable
characters (not ?) and \\ to escape \ itself.

14 years agoFix the setting of module_dirlen for a module path of "/".
Wayne Davison [Fri, 1 Apr 2005 00:18:40 +0000 (00:18 +0000)]
Fix the setting of module_dirlen for a module path of "/".
This fixes a crash in the exclude code when chroot is off.

14 years agoDon't try to tweak read-only dest-dir permissions in list-only mode.
Wayne Davison [Thu, 31 Mar 2005 23:30:03 +0000 (23:30 +0000)]
Don't try to tweak read-only dest-dir permissions in list-only mode.

14 years agoPreparing for release of 2.6.4 v2.6.4
Wayne Davison [Thu, 31 Mar 2005 03:14:34 +0000 (03:14 +0000)]
Preparing for release of 2.6.4

14 years ago- Made the handling of adjacent percents (e.g. "%%b") work like it
Wayne Davison [Thu, 31 Mar 2005 03:01:26 +0000 (03:01 +0000)]
- Made the handling of adjacent percents (e.g. "%%b") work like it
  did in older versions.
- Added a comment for log_format_has().

14 years agoIn log_formatted(), add the newline to the format string so
Wayne Davison [Thu, 31 Mar 2005 01:09:18 +0000 (01:09 +0000)]
In log_formatted(), add the newline to the format string so
that we can call rwrite() instead of rprintf().

14 years agoEnabled width-sizing for %L.
Wayne Davison [Thu, 31 Mar 2005 00:50:58 +0000 (00:50 +0000)]
Enabled width-sizing for %L.

14 years agoUse new log_format_has() function instead of strstr().
Wayne Davison [Thu, 31 Mar 2005 00:21:15 +0000 (00:21 +0000)]
Use new log_format_has() function instead of strstr().

14 years agoAdded log_format_has().
Wayne Davison [Thu, 31 Mar 2005 00:19:13 +0000 (00:19 +0000)]
Added log_format_has().

14 years agoMake sure that we can't scan past the end of the format string.
Wayne Davison [Wed, 30 Mar 2005 23:39:00 +0000 (23:39 +0000)]
Make sure that we can't scan past the end of the format string.

14 years agoOptimized '%f' a little more.
Wayne Davison [Wed, 30 Mar 2005 20:41:11 +0000 (20:41 +0000)]
Optimized '%f' a little more.

14 years agoAllow %i to have a field width.
Wayne Davison [Wed, 30 Mar 2005 20:18:48 +0000 (20:18 +0000)]
Allow %i to have a field width.

14 years ago- Allow the infix field width to start with a '-'.
Wayne Davison [Wed, 30 Mar 2005 19:44:29 +0000 (19:44 +0000)]
- Allow the infix field width to start with a '-'.
- Apply the field width to all escapes, not just numeric ones.

14 years agoMention the numeric field width now possible in the % escapes.
Wayne Davison [Wed, 30 Mar 2005 19:41:51 +0000 (19:41 +0000)]
Mention the numeric field width now possible in the % escapes.

14 years agoMention that the % escapes can now have a numeric infix.
Wayne Davison [Wed, 30 Mar 2005 19:39:11 +0000 (19:39 +0000)]
Mention that the % escapes can now have a numeric infix.

14 years agodropped the "const" from the return value of safe_fname() because
Wayne Davison [Wed, 30 Mar 2005 19:34:20 +0000 (19:34 +0000)]
dropped the "const" from the return value of safe_fname() because
some callers would like to manipulate the string in the buffer.

14 years agoAllow the escapes in the log-format string to contain a numeric
Wayne Davison [Wed, 30 Mar 2005 18:27:09 +0000 (18:27 +0000)]
Allow the escapes in the log-format string to contain a numeric
field width (e.g. %8l %07p).

14 years agoFixed a comment.
Wayne Davison [Wed, 30 Mar 2005 17:31:35 +0000 (17:31 +0000)]
Fixed a comment.

14 years agoMoved a call to output_summary() up a line.
Wayne Davison [Wed, 30 Mar 2005 16:55:11 +0000 (16:55 +0000)]
Moved a call to output_summary() up a line.

14 years agoTweaked a couple sentences.
Wayne Davison [Wed, 30 Mar 2005 16:51:33 +0000 (16:51 +0000)]
Tweaked a couple sentences.

14 years agoChanged %i to output '>' for a local copy.
Wayne Davison [Wed, 30 Mar 2005 16:35:01 +0000 (16:35 +0000)]
Changed %i to output '>' for a local copy.

14 years agoGot rid of a useless extern.
Wayne Davison [Wed, 30 Mar 2005 16:33:41 +0000 (16:33 +0000)]
Got rid of a useless extern.

14 years ago- Fixed a potential crash if the receiver couldn't open the basis file
Wayne Davison [Tue, 29 Mar 2005 22:05:33 +0000 (22:05 +0000)]
- Fixed a potential crash if the receiver couldn't open the basis file
  and the sender (possibly via a batch) tells us to use basis data.
- Enhanced the batched-update skipping message to indicate what phase
  is being skipped.

14 years ago- Warn about missing batched updates the generator wants but the
Wayne Davison [Tue, 29 Mar 2005 19:49:40 +0000 (19:49 +0000)]
- Warn about missing batched updates the generator wants but the
  batch doesn't have.
- Tweaked the "skipping ..." message that is output for already-
  applied batched updates.

14 years agoSplit report() into handle_stats() and output_report() so that (1)
Wayne Davison [Tue, 29 Mar 2005 11:11:35 +0000 (11:11 +0000)]
Split report() into handle_stats() and output_report() so that (1)
the report happens after all --delete-after activity has ceased when
pulling, and (2) a batch-file created when pushing gets the stats
values written out prior to any end-of-run keep-alive packets.

14 years agoChanged maybe_emit_filelist_progress() and emit_filelist_progress()
Wayne Davison [Tue, 29 Mar 2005 10:57:31 +0000 (10:57 +0000)]
Changed maybe_emit_filelist_progress() and emit_filelist_progress()
to take an integer count instead of a struct file_list so that we
can pass in a file-list-count offset for delete's separate calls
to send_directory().

14 years agoTurn off do_progress during the file-update phase so that
Wayne Davison [Tue, 29 Mar 2005 10:41:26 +0000 (10:41 +0000)]
Turn off do_progress during the file-update phase so that
--delete-during and --fuzzy don't get any superfluous
" N files..." lines output in the middle of the processing.

14 years agoPreparing for release of 2.6.4pre4 v2.6.4pre4
Wayne Davison [Tue, 29 Mar 2005 06:18:24 +0000 (06:18 +0000)]
Preparing for release of 2.6.4pre4

14 years agoImproved several items and preparing for final 2.6.4 release.
Wayne Davison [Tue, 29 Mar 2005 06:15:47 +0000 (06:15 +0000)]
Improved several items and preparing for final 2.6.4 release.

14 years agoAdd data for 2.6.4.
Wayne Davison [Tue, 29 Mar 2005 06:15:08 +0000 (06:15 +0000)]
Add data for 2.6.4.

14 years agoIf --delay-updates wasn't specified, write out two -1 indexes at
Wayne Davison [Tue, 29 Mar 2005 01:04:20 +0000 (01:04 +0000)]
If --delay-updates wasn't specified, write out two -1 indexes at
the same time to avoid a useless round-trip delay for an empty
delay-updates phase.

14 years agoAdded --protocol and made some batch-file improvements.
Wayne Davison [Mon, 28 Mar 2005 20:54:06 +0000 (20:54 +0000)]
Added --protocol and made some batch-file improvements.

14 years agoAdded --protocol to --help output.
Wayne Davison [Mon, 28 Mar 2005 20:40:25 +0000 (20:40 +0000)]
Added --protocol to --help output.

14 years agoSome temporary code for 2.6.4pre3 compatibility (which can be
Wayne Davison [Mon, 28 Mar 2005 17:29:27 +0000 (17:29 +0000)]
Some temporary code for 2.6.4pre3 compatibility (which can be
safely removed for the next release).

14 years agoChaging ITEM_UNUSED back to ITEM_DUMMY_BIT for now -- we'll need
Wayne Davison [Mon, 28 Mar 2005 17:08:47 +0000 (17:08 +0000)]
Chaging ITEM_UNUSED back to ITEM_DUMMY_BIT for now -- we'll need
to get rid of it later on.

14 years agoChanged ITEM_DUMMY_BIT to ITEM_UNUSED.
Wayne Davison [Sun, 27 Mar 2005 05:58:40 +0000 (05:58 +0000)]
Changed ITEM_DUMMY_BIT to ITEM_UNUSED.

14 years agoGot rid of temporary code that detected and rejected older 2.6.4
Wayne Davison [Sun, 27 Mar 2005 05:58:38 +0000 (05:58 +0000)]
Got rid of temporary code that detected and rejected older 2.6.4
pre-release versions.

14 years agoGot rid of temporary code that set the ITEM_DUMMY_BIT.
Wayne Davison [Sun, 27 Mar 2005 05:58:36 +0000 (05:58 +0000)]
Got rid of temporary code that set the ITEM_DUMMY_BIT.

14 years ago--compress is now saved in a batch's option-state flags.
Wayne Davison [Sun, 27 Mar 2005 05:49:24 +0000 (05:49 +0000)]
--compress is now saved in a batch's option-state flags.

14 years agoDecided that the '<' and '>' output in the %i format were
Wayne Davison [Sun, 27 Mar 2005 05:32:36 +0000 (05:32 +0000)]
Decided that the '<' and '>' output in the %i format were
the opposite of what they should be.

14 years ago- Fixed a bug in the saving of the --dirs option's state.
Wayne Davison [Sun, 27 Mar 2005 05:02:49 +0000 (05:02 +0000)]
- Fixed a bug in the saving of the --dirs option's state.
- Added the saving of the --compress option's state.
- Deal with the xfer_dirs var in a better way for pre-29 batches.

14 years agoMention that --dry-run no longer conflicts with the batch options.
Wayne Davison [Fri, 25 Mar 2005 16:45:58 +0000 (16:45 +0000)]
Mention that --dry-run no longer conflicts with the batch options.

14 years ago- Don't complain if --dry-run is specified with --read-batch
Wayne Davison [Thu, 24 Mar 2005 16:41:16 +0000 (16:41 +0000)]
- Don't complain if --dry-run is specified with --read-batch
  or --write-batch.
- If --write-batch is combined with --dry-run, just disable
  --write-batch (that avoids trying to create a batch file
  and tells the user what would be transferred).

14 years agoIf --dry-run is enabled with --read-batch, we must discard the
Wayne Davison [Thu, 24 Mar 2005 16:38:34 +0000 (16:38 +0000)]
If --dry-run is enabled with --read-batch, we must discard the
transfer data.

14 years agoList /etc instead of /, but only if it exists and is readable.
Wayne Davison [Wed, 23 Mar 2005 16:04:17 +0000 (16:04 +0000)]
List /etc instead of /, but only if it exists and is readable.

14 years agoMention when we run fakeroot.
Wayne Davison [Fri, 18 Mar 2005 02:10:34 +0000 (02:10 +0000)]
Mention when we run fakeroot.

14 years agoIf we're not root and the "fakeroot" command is available, use it
Wayne Davison [Fri, 18 Mar 2005 02:07:25 +0000 (02:07 +0000)]
If we're not root and the "fakeroot" command is available, use it
to re-run the script while pretending to be root.

14 years agoSet a maximum distance-measure that find_fuzzy() will accept.
Wayne Davison [Thu, 17 Mar 2005 08:59:48 +0000 (08:59 +0000)]
Set a maximum distance-measure that find_fuzzy() will accept.

14 years agoFixed a problem with the stripping of the .bak/.old/.orig suffixes
Wayne Davison [Thu, 17 Mar 2005 08:45:36 +0000 (08:45 +0000)]
Fixed a problem with the stripping of the .bak/.old/.orig suffixes
in find_filename_suffix().

14 years agoFixed the -r kluge sent for pre-2.6.4 --list-only support.
Wayne Davison [Thu, 17 Mar 2005 00:52:33 +0000 (00:52 +0000)]
Fixed the -r kluge sent for pre-2.6.4 --list-only support.