rsync/rsync.git
20 years agoAdded stringjoin() and pathjoin().
Wayne Davison [Tue, 20 Jan 2004 04:43:49 +0000 (04:43 +0000)]
Added stringjoin() and pathjoin().

20 years agoEliminate unneeded strlen after strlcpy.
J.W. Schultz [Tue, 20 Jan 2004 03:37:04 +0000 (03:37 +0000)]
Eliminate unneeded strlen after strlcpy.

20 years agoMake sure that strlcpy() returns the right value when the bufsize is 0.
Wayne Davison [Tue, 20 Jan 2004 00:59:26 +0000 (00:59 +0000)]
Make sure that strlcpy() returns the right value when the bufsize is 0.

20 years agoFixed a bug in strlcat() where it would not properly detect a no-change
Wayne Davison [Tue, 20 Jan 2004 00:29:49 +0000 (00:29 +0000)]
Fixed a bug in strlcat() where it would not properly detect a no-change
condition if called with an initial string longer than the specified
size limit (due to an unsigned var's inability to go negative).

20 years agoMake sure that the batch-prefix string isn't overly long (avoiding
Wayne Davison [Mon, 19 Jan 2004 23:33:02 +0000 (23:33 +0000)]
Make sure that the batch-prefix string isn't overly long (avoiding
a potential problem in the batch-filename code).

20 years agoActually, let's avoid snprintf() where possible.
Wayne Davison [Mon, 19 Jan 2004 23:28:57 +0000 (23:28 +0000)]
Actually, let's avoid snprintf() where possible.

20 years agoUse snprintf() instead of strlcpy()+strlcat().
Wayne Davison [Sun, 18 Jan 2004 21:22:49 +0000 (21:22 +0000)]
Use snprintf() instead of strlcpy()+strlcat().

20 years agoA few minor string optimizations and a new complaint if we try to
Wayne Davison [Sat, 17 Jan 2004 19:45:26 +0000 (19:45 +0000)]
A few minor string optimizations and a new complaint if we try to
send a truncated filename.

20 years agoGot rid of an extra call to io_start_buffering_out().
Wayne Davison [Sat, 17 Jan 2004 05:04:04 +0000 (05:04 +0000)]
Got rid of an extra call to io_start_buffering_out().

20 years agoUnified the externs.
Wayne Davison [Sat, 17 Jan 2004 04:58:24 +0000 (04:58 +0000)]
Unified the externs.

20 years agoA fix for --cvs-exclude.
Wayne Davison [Sat, 17 Jan 2004 01:23:41 +0000 (01:23 +0000)]
A fix for --cvs-exclude.

20 years agoItems read from a per-directory .cvsignore file should be added to the
Wayne Davison [Sat, 17 Jan 2004 01:16:49 +0000 (01:16 +0000)]
Items read from a per-directory .cvsignore file should be added to the
local_exclude_list, not the exclude_list.

20 years agoOutput an error before we die if we get a wacko message from the receiver.
Wayne Davison [Fri, 16 Jan 2004 16:31:47 +0000 (16:31 +0000)]
Output an error before we die if we get a wacko message from the receiver.

20 years agoClarified the --delete-after descriptions.
Wayne Davison [Thu, 15 Jan 2004 17:45:53 +0000 (17:45 +0000)]
Clarified the --delete-after descriptions.

20 years agoClarified the --delete-after help text.
Wayne Davison [Thu, 15 Jan 2004 17:43:34 +0000 (17:43 +0000)]
Clarified the --delete-after help text.

20 years agoIf we're a daemon talking with a pre-28 protocol, use FERROR for a
Wayne Davison [Thu, 15 Jan 2004 08:56:33 +0000 (08:56 +0000)]
If we're a daemon talking with a pre-28 protocol, use FERROR for a
vanished-file message (not FINFO).

20 years agoIn two places: needed to use protocol_version, not PROTOCOL_VERSION.
Wayne Davison [Thu, 15 Jan 2004 08:39:25 +0000 (08:39 +0000)]
In two places: needed to use protocol_version, not PROTOCOL_VERSION.

20 years agoChanged version to "2.6.1cvs".
Wayne Davison [Thu, 15 Jan 2004 08:16:40 +0000 (08:16 +0000)]
Changed version to "2.6.1cvs".

20 years agoImproved receiver/generator communications.
Wayne Davison [Thu, 15 Jan 2004 07:51:37 +0000 (07:51 +0000)]
Improved receiver/generator communications.

20 years ago- Define FULL_FLUSH and NORMAL_FLUSH.
Wayne Davison [Thu, 15 Jan 2004 07:42:27 +0000 (07:42 +0000)]
- Define FULL_FLUSH and NORMAL_FLUSH.
- Got rid of FNONE from logcode enum.
- Created msgcode enum with MSG_DATA, MSG_REDO, MSG_DONE, etc.

20 years ago- Got rid of the f_gen pipe.
Wayne Davison [Thu, 15 Jan 2004 07:42:25 +0000 (07:42 +0000)]
- Got rid of the f_gen pipe.
- Use new send_msg() call with MSG_DONE or MSG_REDO.

20 years ago- Call io_flush() with its new FULL_FLUSH/NORMAL_FLUSH arg.
Wayne Davison [Thu, 15 Jan 2004 07:42:21 +0000 (07:42 +0000)]
- Call io_flush() with its new FULL_FLUSH/NORMAL_FLUSH arg.
- Got rid of the recv_pipe between the receiver and the generator.
- Use new send_msg(MSG_DONE,...) call to signal the generator
  when the receiver is done.
- Use the new get_redo_num() call in the generator to read the
  final -1 from the receiver.

20 years ago- Moved err_list* stuff to io.c and renamed them to msg_list*.
Wayne Davison [Thu, 15 Jan 2004 07:42:19 +0000 (07:42 +0000)]
- Moved err_list* stuff to io.c and renamed them to msg_list*.
- Call the new send_msg() function to send a remote log message.

20 years ago- Moved the err_list* stuff here from log.c and renamed them msg_list*.
Wayne Davison [Thu, 15 Jan 2004 07:42:17 +0000 (07:42 +0000)]
- Moved the err_list* stuff here from log.c and renamed them msg_list*.
- Renamed log_error_fd to msg_fd_out, and io_error_fd to msg_fd_in.
- Added a redo_list* to keep track of any redo items that come over
  the message channel from the receiver to the generator.
- The io_flush() and msg_list_push() functions now take an arg:
  NORMAL_FLUSH or FULL_FLUSH.  The latter ensures that we don't lose
  any log messages that haven't gotten written from the msg_list.
- Added a send_msg() function that log.c calls to send any remote
  log messages.  It is also called to send the new MSG_REDO and
  MSG_DONE messages.

20 years agoWe now call get_redo_num() instead of reading f_recv (which was removed).
Wayne Davison [Thu, 15 Jan 2004 07:42:15 +0000 (07:42 +0000)]
We now call get_redo_num() instead of reading f_recv (which was removed).

20 years agoCall io_flush() with its new FULL_FLUSH option.
Wayne Davison [Thu, 15 Jan 2004 07:42:12 +0000 (07:42 +0000)]
Call io_flush() with its new FULL_FLUSH option.

20 years agoIf the do_lchown() fails and we weren't trying to change the user,
Wayne Davison [Thu, 15 Jan 2004 07:08:00 +0000 (07:08 +0000)]
If the do_lchown() fails and we weren't trying to change the user,
the error now says that chgrp failed, not chown.

20 years agoFixed a bug when a non-root user copies a GID=0 file with -g set
Wayne Davison [Thu, 15 Jan 2004 07:06:45 +0000 (07:06 +0000)]
Fixed a bug when a non-root user copies a GID=0 file with -g set
and no permissions to set group 0.

20 years agoThe HAS_INODE_DATA flag is no longer a part of the LIVE_FLAGS define.
Wayne Davison [Thu, 15 Jan 2004 02:25:38 +0000 (02:25 +0000)]
The HAS_INODE_DATA flag is no longer a part of the LIVE_FLAGS define.

20 years agoExplain the --progress output.
Wayne Davison [Tue, 13 Jan 2004 18:50:40 +0000 (18:50 +0000)]
Explain the --progress output.

20 years agoGive the user an idea of how far along in the transfer we are
Wayne Davison [Tue, 13 Jan 2004 18:25:03 +0000 (18:25 +0000)]
Give the user an idea of how far along in the transfer we are
when using --progress.

20 years agoSet stats.current_file_index.
Wayne Davison [Tue, 13 Jan 2004 18:22:43 +0000 (18:22 +0000)]
Set stats.current_file_index.

20 years agoAdded current_file_index.
Wayne Davison [Tue, 13 Jan 2004 18:22:13 +0000 (18:22 +0000)]
Added current_file_index.

20 years agoMerged a couple more f != -1 if sections.
Wayne Davison [Tue, 13 Jan 2004 06:27:30 +0000 (06:27 +0000)]
Merged a couple more f != -1 if sections.

20 years agoMoved a couple externs.
Wayne Davison [Tue, 13 Jan 2004 05:13:57 +0000 (05:13 +0000)]
Moved a couple externs.

20 years agoMerged some "if (f != -1)" checks, put the clean_flist() call in the
Wayne Davison [Tue, 13 Jan 2004 05:11:56 +0000 (05:11 +0000)]
Merged some "if (f != -1)" checks, put the clean_flist() call in the
right spot, and moved some externs.

20 years agoRelocated some externs.
Wayne Davison [Tue, 13 Jan 2004 04:50:45 +0000 (04:50 +0000)]
Relocated some externs.

20 years agoMore hard-link stuff.
Wayne Davison [Mon, 12 Jan 2004 03:59:53 +0000 (03:59 +0000)]
More hard-link stuff.

20 years agoChanged the hard-link check.
Wayne Davison [Mon, 12 Jan 2004 03:50:07 +0000 (03:50 +0000)]
Changed the hard-link check.

20 years agoUpdated to perform a much more efficient hlink algorithm that doesn't
Wayne Davison [Mon, 12 Jan 2004 03:49:47 +0000 (03:49 +0000)]
Updated to perform a much more efficient hlink algorithm that doesn't
require any binary searching of hlink data.

20 years agoAdded F_HEAD and F_NEXT defines.
Wayne Davison [Mon, 12 Jan 2004 03:48:43 +0000 (03:48 +0000)]
Added F_HEAD and F_NEXT defines.

20 years agoMention more of the latest changes.
Wayne Davison [Sun, 11 Jan 2004 22:39:34 +0000 (22:39 +0000)]
Mention more of the latest changes.

20 years agoMore cleanup, fixing an error in the fatal-message strings that tried
Wayne Davison [Sun, 11 Jan 2004 22:11:25 +0000 (22:11 +0000)]
More cleanup, fixing an error in the fatal-message strings that tried
to output the "filename" value when it might not be initialized.

20 years agoUpdated a questioning comment.
Wayne Davison [Sun, 11 Jan 2004 08:17:19 +0000 (08:17 +0000)]
Updated a questioning comment.

20 years agoUse the new dev+inode union in the flist_struct.
Wayne Davison [Sun, 11 Jan 2004 07:56:14 +0000 (07:56 +0000)]
Use the new dev+inode union in the flist_struct.

20 years agoMoved the inode & dev items out of the flist_struct. Based on a
Wayne Davison [Sun, 11 Jan 2004 07:55:53 +0000 (07:55 +0000)]
Moved the inode & dev items out of the flist_struct.  Based on a
preliminary patch by J.W.

20 years agoUse the new union-ified flist_struct values.
Wayne Davison [Sun, 11 Jan 2004 07:28:45 +0000 (07:28 +0000)]
Use the new union-ified flist_struct values.

20 years agoPut 3 flist_struct items into a union: rdev, sum, and link (all
Wayne Davison [Sun, 11 Jan 2004 07:28:23 +0000 (07:28 +0000)]
Put 3 flist_struct items into a union: rdev, sum, and link (all
mutually exclusive based on mode).  Based on an idea by J.W.

20 years agoA slightly better reset mechanism for send_file_entry() and
Wayne Davison [Sat, 10 Jan 2004 22:13:19 +0000 (22:13 +0000)]
A slightly better reset mechanism for send_file_entry() and
receive_file_entry().

20 years agoUse the new LIVE_FLAGS define to fix a potential flag problem.
Wayne Davison [Sat, 10 Jan 2004 21:10:04 +0000 (21:10 +0000)]
Use the new LIVE_FLAGS define to fix a potential flag problem.

20 years agoAdded LIVE_FLAGS define.
Wayne Davison [Sat, 10 Jan 2004 21:09:06 +0000 (21:09 +0000)]
Added LIVE_FLAGS define.

20 years agoChanged the call to write_batch_csum_info() to the latest syntax.
Wayne Davison [Sat, 10 Jan 2004 20:16:18 +0000 (20:16 +0000)]
Changed the call to write_batch_csum_info() to the latest syntax.

20 years agoMassive rewrite to replace the file-list read/write code with calls to
Wayne Davison [Sat, 10 Jan 2004 20:16:13 +0000 (20:16 +0000)]
Massive rewrite to replace the file-list read/write code with calls to
send_file_entry() and receive_file_entry().  Other cleanups too.

20 years agoAllow the send_file_entry() and receive_file_entry() routines to be
Wayne Davison [Sat, 10 Jan 2004 20:15:55 +0000 (20:15 +0000)]
Allow the send_file_entry() and receive_file_entry() routines to be
called by the batch processor.

20 years agoSupport the recent changes in the flist_struct.
Wayne Davison [Sat, 10 Jan 2004 08:39:57 +0000 (08:39 +0000)]
Support the recent changes in the flist_struct.

20 years agoGot rid of tmpsum[].
Wayne Davison [Fri, 9 Jan 2004 18:35:41 +0000 (18:35 +0000)]
Got rid of tmpsum[].

20 years agoNote latest changes.
Wayne Davison [Fri, 9 Jan 2004 18:27:15 +0000 (18:27 +0000)]
Note latest changes.

20 years ago- Never allocate a useless sum array for a non-regular flist item.
Wayne Davison [Fri, 9 Jan 2004 18:19:32 +0000 (18:19 +0000)]
- Never allocate a useless sum array for a non-regular flist item.
- Don't transmit the sum array for non-regular files (p28 onward).
- Tweaked a couple comments.

20 years agoA couple minor code improvements.
Wayne Davison [Fri, 9 Jan 2004 18:10:31 +0000 (18:10 +0000)]
A couple minor code improvements.

20 years agoAdded an extra comment.
Wayne Davison [Fri, 9 Jan 2004 16:53:55 +0000 (16:53 +0000)]
Added an extra comment.

20 years agoAdded a few more devices to the devices.test to hopefully
J.W. Schultz [Fri, 9 Jan 2004 14:02:00 +0000 (14:02 +0000)]
Added a few more devices to the devices.test to hopefully
test same, same high and different device numbers.

20 years agoFinished the last_* renameing and fixed a cast..
J.W. Schultz [Fri, 9 Jan 2004 13:55:56 +0000 (13:55 +0000)]
Finished the last_* renameing and fixed a cast..

20 years agoUpdated the send_file_name() and receive_file_entry() functions to make
Wayne Davison [Thu, 8 Jan 2004 22:45:12 +0000 (22:45 +0000)]
Updated the send_file_name() and receive_file_entry() functions to make
some of the variable names more logical in the new program flow.  Also
improved the setting of the new last_rdev value that only remembers the
high 3 bytes.

20 years agoChanged the OLD_SAME_RDEV define to SAME_RDEV_pre28.
Wayne Davison [Thu, 8 Jan 2004 22:45:07 +0000 (22:45 +0000)]
Changed the OLD_SAME_RDEV define to SAME_RDEV_pre28.

20 years agoAdded explicit call to AC_PROG_EGREP.
Wayne Davison [Thu, 8 Jan 2004 18:33:38 +0000 (18:33 +0000)]
Added explicit call to AC_PROG_EGREP.

20 years agoOptimized the sending of dev+inode data for hard-linking support
Wayne Davison [Thu, 8 Jan 2004 11:29:50 +0000 (11:29 +0000)]
Optimized the sending of dev+inode data for hard-linking support
(but only when speaking protocol 28).

20 years ago- Added flags HAS_INODE_DATA and SAME_DEV.
Wayne Davison [Thu, 8 Jan 2004 11:28:07 +0000 (11:28 +0000)]
- Added flags HAS_INODE_DATA and SAME_DEV.
- Changed PROTOCOL_VERSION to 28 now.

20 years agoImplemented an extended-flag-byte in the transfer protocol by moving
Wayne Davison [Thu, 8 Jan 2004 11:08:30 +0000 (11:08 +0000)]
Implemented an extended-flag-byte in the transfer protocol by moving
the SAME_RDEV bit into the next 8 bits and adding an EXTENDED_FLAG
bit.  Also modified the meaning of this bit for protocol 28 and above:
SAME_HIGH_RDEV indicates that all but the lowest byte matches.  The
older protocols get OLD_SAME_RDEV set (with the old semantics) for
compatibility.  Note also that protocol 28 isn't turned on yet.

20 years agoAdded EXTENDED_FLAGS and SAME_HIGH_RDEV. Changed SAME_RDEV to OLD_SAME_RDEV.
Wayne Davison [Thu, 8 Jan 2004 11:08:30 +0000 (11:08 +0000)]
Added EXTENDED_FLAGS and SAME_HIGH_RDEV. Changed SAME_RDEV to OLD_SAME_RDEV.

20 years agoPass the $srcdir/wildtest.txt file on the command-line to wildtest.
Wayne Davison [Thu, 8 Jan 2004 09:00:14 +0000 (09:00 +0000)]
Pass the $srcdir/wildtest.txt file on the command-line to wildtest.

20 years agoWe now take a TESTFILE arg on the command-line.
Wayne Davison [Thu, 8 Jan 2004 08:59:52 +0000 (08:59 +0000)]
We now take a TESTFILE arg on the command-line.

20 years agoImproved the code in send_file_entry() and receive_file_entry() to
Wayne Davison [Thu, 8 Jan 2004 04:56:27 +0000 (04:56 +0000)]
Improved the code in send_file_entry() and receive_file_entry() to
make it a little easier to read and a little more optimized.  Also
made the flag size in our function calls match the size of the flags
in the flist_struct.  These changes will make it easier for me to
optimize the sending of the hard-link data in a future commit.

20 years agoMade the flags in the file_struct a little smaller.
Wayne Davison [Thu, 8 Jan 2004 04:53:16 +0000 (04:53 +0000)]
Made the flags in the file_struct a little smaller.

20 years ago- No need to check S_ISREG() on files already in the hlink_list.
Wayne Davison [Thu, 8 Jan 2004 03:17:07 +0000 (03:17 +0000)]
- No need to check S_ISREG() on files already in the hlink_list.
- The realloc in init_hard_links() might fail if the list is empty
  (which can happen if there are no regular files in the transfer).
  I changed the code to free the list in that case.

20 years agoOld-protocols.
Wayne Davison [Thu, 8 Jan 2004 00:47:17 +0000 (00:47 +0000)]
Old-protocols.

20 years agoGot rid of support for protocols 17 and 18.
Wayne Davison [Thu, 8 Jan 2004 00:45:41 +0000 (00:45 +0000)]
Got rid of support for protocols 17 and 18.

20 years agoNoted hardlink memory footprint reduction.
J.W. Schultz [Wed, 7 Jan 2004 11:38:23 +0000 (11:38 +0000)]
Noted hardlink memory footprint reduction.

20 years agoCosmetic changes to file_compare() for readability and to
J.W. Schultz [Wed, 7 Jan 2004 09:11:56 +0000 (09:11 +0000)]
Cosmetic changes to file_compare() for readability and to
match hlink_compare()

20 years agoChange hlink_list so we only have a list of pointers to
J.W. Schultz [Wed, 7 Jan 2004 09:06:37 +0000 (09:06 +0000)]
Change hlink_list so we only have a list of pointers to
file_struct instead of a copy of the file_struct list.

20 years agoBatch-mode fix.
Wayne Davison [Tue, 6 Jan 2004 05:35:55 +0000 (05:35 +0000)]
Batch-mode fix.

20 years agoFixed a byte-order problem for batch-file processing on big-endian
Wayne Davison [Tue, 6 Jan 2004 05:33:02 +0000 (05:33 +0000)]
Fixed a byte-order problem for batch-file processing on big-endian
systems (reported by Jay Fenlason).

20 years agoSome whitespace tweaks.
Wayne Davison [Mon, 5 Jan 2004 03:57:15 +0000 (03:57 +0000)]
Some whitespace tweaks.

20 years agoA few formatting tweaks.
Wayne Davison [Mon, 5 Jan 2004 03:56:19 +0000 (03:56 +0000)]
A few formatting tweaks.

20 years agoUse new recv_generator() calling syntax.
Wayne Davison [Sun, 4 Jan 2004 19:18:37 +0000 (19:18 +0000)]
Use new recv_generator() calling syntax.

20 years agoChanged recv_generator() to take a "struct file_struct *" instead of
Wayne Davison [Sun, 4 Jan 2004 18:43:18 +0000 (18:43 +0000)]
Changed recv_generator() to take a "struct file_struct *" instead of
a "struct file_list *", which allows us to call it with a copy of the
current file_struct entry that we've modified.  This is better than
modifying the entry directly because the latter causes the shared
memory between the generator and the receiver to slowly vanish as
the in-memory changes happen.

20 years agoTry using "id -un" before "whoami" since the latter doesn't exist on
Wayne Davison [Sun, 4 Jan 2004 07:07:08 +0000 (07:07 +0000)]
Try using "id -un" before "whoami" since the latter doesn't exist on
SunOS systems (and I hope the former works -- we'll see).

20 years agoSilence a compiler warning on Sun OS systems.
Wayne Davison [Sun, 4 Jan 2004 07:06:05 +0000 (07:06 +0000)]
Silence a compiler warning on Sun OS systems.

20 years agoImproved the text.
Wayne Davison [Sat, 3 Jan 2004 20:50:51 +0000 (20:50 +0000)]
Improved the text.

20 years agoFixed autom4te*.cache entry.
Wayne Davison [Sat, 3 Jan 2004 19:49:52 +0000 (19:49 +0000)]
Fixed autom4te*.cache entry.

20 years agoMention more INTERNALS work.
Wayne Davison [Sat, 3 Jan 2004 19:28:17 +0000 (19:28 +0000)]
Mention more INTERNALS work.

20 years agoMore variable-size cleanup.
Wayne Davison [Sat, 3 Jan 2004 19:28:03 +0000 (19:28 +0000)]
More variable-size cleanup.

20 years agoMade the types used in the sum_buf and sum_struct structures consistent
Wayne Davison [Sat, 3 Jan 2004 18:37:41 +0000 (18:37 +0000)]
Made the types used in the sum_buf and sum_struct structures consistent
with the variables in the code that manipulates these values.

20 years agoCreate GID_NONE for use gid test readability.
J.W. Schultz [Sat, 3 Jan 2004 13:17:32 +0000 (13:17 +0000)]
Create GID_NONE for use gid test readability.

20 years agoFixed a name.
Wayne Davison [Sat, 3 Jan 2004 09:48:52 +0000 (09:48 +0000)]
Fixed a name.

20 years agoFixed a size check in the new code.
Wayne Davison [Sat, 3 Jan 2004 09:44:20 +0000 (09:44 +0000)]
Fixed a size check in the new code.

20 years agoDocument the latest changes.
Wayne Davison [Sat, 3 Jan 2004 09:36:30 +0000 (09:36 +0000)]
Document the latest changes.

20 years agoAdded support for "Basic" authentication to our proxy, based on a
Wayne Davison [Sat, 3 Jan 2004 09:19:06 +0000 (09:19 +0000)]
Added support for "Basic" authentication to our proxy, based on a
patch by Bardur Arantsson.

20 years agoMade base64_encode() non-static.
Wayne Davison [Sat, 3 Jan 2004 08:53:36 +0000 (08:53 +0000)]
Made base64_encode() non-static.

20 years agoSilence a compiler warning.
Wayne Davison [Sat, 3 Jan 2004 01:12:10 +0000 (01:12 +0000)]
Silence a compiler warning.

20 years agoAdded missing '[]'s to AC_CHECK_FUNCS() calls so that this works
Wayne Davison [Sat, 3 Jan 2004 00:40:55 +0000 (00:40 +0000)]
Added missing '[]'s to AC_CHECK_FUNCS() calls so that this works
with autoconf 2.58 and above.