Commit Graph

  • 649f874292 Moved some of the delete code from receiver.c to here and implemented the handling for delete_during (which is called from the generator). Wayne Davison 2005-01-24 01:41:42 +00:00
  • 90fdd89a0b Mention the --omit-dir-times option. Wayne Davison 2005-01-24 00:56:37 +00:00
  • 54e66f1d59 Document --omit-dir-times. Wayne Davison 2005-01-24 00:51:38 +00:00
  • 82471e68a8 Honor the new omit_dir_times var. Wayne Davison 2005-01-24 00:51:34 +00:00
  • 20fb7b9175 Added --omit-dir-times (-O). Wayne Davison 2005-01-24 00:51:32 +00:00
  • 859fdaad45 Mentioned the --dirs and --list-only options. Wayne Davison 2005-01-24 00:34:53 +00:00
  • 09ed309996 - Changed --keep-dirs (-k) into --dirs (-d). - Document --list-only. Wayne Davison 2005-01-24 00:17:42 +00:00
  • 65e4cda059 - Changed --keep-dirs (-k) into --dirs (-d). - Renamed keep_dirs to xfer_dirs. - Mention --list-only in the --help. Wayne Davison 2005-01-24 00:17:39 +00:00
  • 7e037c4226 Renamed keep_dirs, "xfer_dirs". Wayne Davison 2005-01-24 00:17:36 +00:00
  • f636c38440 Document the new --keep-dirs option. Wayne Davison 2005-01-23 16:49:09 +00:00
  • b98f040ef2 Go back to requiring -r for deletes to happen. Wayne Davison 2005-01-23 07:27:01 +00:00
  • e57211c544 When an option defined with POPT_ARG_VAL is being refused, we must change it to POPT_ARG_NONE or it won't be refused. Wayne Davison 2005-01-23 01:25:36 +00:00
  • 48a1ff0d45 - Use the new delete_before var. - Check the "keep_dirs" var instead of "recurse" to decide if delete is appropriate. Wayne Davison 2005-01-22 22:48:40 +00:00
  • 032dcf74b0 Check the "keep_dirs" var instead of "recurse" to decide if delete-after is appropriate. Wayne Davison 2005-01-22 22:48:37 +00:00
  • 51d4839861 - We now set a delete_before variable for the --delete option. - Reordered some of the options in the --help list. - Made the popt-handling of the delete options simpler. - Parse the new --keep-dirs option. - Parse the new --list-only option. - Set recurse to -1 (infinite) for the -r and -a options. - Set keep_dirs for --list-only. Wayne Davison 2005-01-22 22:48:34 +00:00
  • 9f125ea7c1 - Implemented a new handling of the recurse variable: if it is < 0, it specifies infinite recursion, otherwise it contains a count of how many levels of recursion we will allow. - If the user is processing/listing a non-recursive directory/ arg, send the contents. Wayne Davison 2005-01-22 22:48:31 +00:00
  • 9bcb25958d - Mustn't override a user-specified list_only value. - Moved the computing of the options for the server down until after we've figured out the protocol_version for the transfer. - If we're talking to a protocol-29 server in list-only mode, force the new --list-only mode (which avoids the -r with --exclude="/*/*" kluge). - Output the options we sent to the daemon if verbose > 1. Wayne Davison 2005-01-22 22:48:27 +00:00
  • 866925bfb7 Changed the order of a few of the options in the options summary. Wayne Davison 2005-01-22 20:32:03 +00:00
  • f8b3c05311 Mustn't override a user-specified list_only value. Wayne Davison 2005-01-22 20:28:17 +00:00
  • 353f272434 Changed a comment. Wayne Davison 2005-01-22 20:27:22 +00:00
  • 3d54c6ec46 Moved the /proc/mounts literal string nearer to the top. Wayne Davison 2005-01-22 08:11:43 +00:00
  • 71e27c463d - Changed XMIT_TOP_DIR to XMIT_DEL_START. - Changed FLAG_TOP_DIR to FLAG_DEL_START. Wayne Davison 2005-01-21 00:34:37 +00:00
  • 8c48382049 Don't mention when the file-list is first allocated, just when it expands. Wayne Davison 2005-01-20 23:57:35 +00:00
  • 11781089d1 In delete_file(): got rid of DEL_NO_RECURSE code and added DEL_RECURSE code. Improved the verbose output of a recursively deleted directory when --dry-run is specified. Wayne Davison 2005-01-20 23:51:57 +00:00
  • d97fd43a72 Decided that delete_in_dir()'s call to delete_file() would be better using DEL_RECURSE. Wayne Davison 2005-01-20 23:51:55 +00:00
  • f5ea4b3b39 Changed DEL_NO_RECURSE to DEL_RECURSE. Wayne Davison 2005-01-20 23:51:52 +00:00
  • f227ffe4b9 Moved the dry_run check in set_perms() so that it only triggers if we don't have a stat-buffer (and returns 1, not 0). Wayne Davison 2005-01-20 23:01:34 +00:00
  • 027428eb1d Got rid of the erroneous --dry-run short-circuit in the directory handling. Wayne Davison 2005-01-20 22:59:54 +00:00
  • 15778afbdc In set_modtime(), the verbose message should be output even if --dry-run was specified. Wayne Davison 2005-01-20 22:58:30 +00:00
  • d06f63287e - Got rid of the uint64 typedef. - Changed the dev/inode vars to use int64. Wayne Davison 2005-01-20 22:43:32 +00:00
  • 1490812ab2 Switched uint64 vars over to int64. Wayne Davison 2005-01-20 22:42:47 +00:00
  • 707415d4fc Use int64 instead of uint64. Wayne Davison 2005-01-20 22:37:38 +00:00
  • 9cea6ef1b6 Improved a confusing sentence in the description of subcomponent include/exclude matching. Wayne Davison 2005-01-20 19:31:42 +00:00
  • 48ea74bf01 When outputting the file list created for the delete pass, identify it as such. Wayne Davison 2005-01-20 00:53:15 +00:00
  • 026deaf759 Added an example of running mnt-excl remotely via ssh for a pull. Wayne Davison 2005-01-19 21:07:09 +00:00
  • 4c59971194 A perl script that outputs excludes for all mount points that affect the supplied source dir (or all mount points if the arg is omitted). The excludes are appropriately anchored relative to the supplied dir, and honor rsync's trailing-slash idiom. Wayne Davison 2005-01-19 21:00:26 +00:00
  • fd8571c2c9 Mention a change in the verbose output when deleting directories. Wayne Davison 2005-01-19 20:32:25 +00:00
  • dd096ae080 - Updated delete_file() to take a flag arg that makes it more flexible. Optimized away a stat() call since the caller knows if the item to delete is a directory or not. - Tweaked delete_file()'s "deleting ..." messages to include a trailing slash when we're deleting a directory. - No need for conditional code around do_lstat() anymore. Wayne Davison 2005-01-19 20:11:03 +00:00
  • eb1accaa91 Define the new DEL_* flags. Wayne Davison 2005-01-19 20:11:00 +00:00
  • 7e38410e47 - Moved write_sum_head() to io.c. - Use the new delete_file() calling syntax. Wayne Davison 2005-01-19 20:10:57 +00:00
  • d336388403 Got rid of delete_one() since we can (and do) use the updated delete_file() call now. Wayne Davison 2005-01-19 20:10:53 +00:00
  • c207d7ec62 Moved write_sum_head() here from generator.c. Wayne Davison 2005-01-19 20:10:50 +00:00
  • 018b28328c No need for conditional code around do_lstat() anymore. Wayne Davison 2005-01-19 19:30:29 +00:00
  • 0957a7463e When SUPPORT_LINKS is not defined, we now go ahead and provide the do_lstat() function and just have it call do_stat(). Wayne Davison 2005-01-19 19:29:20 +00:00
  • c72f5bd9c4 A slight simplification to the no-implied-dirs change. Wayne Davison 2005-01-19 00:34:13 +00:00
  • 0f57446da7 Use the new "keep_dirs" var instead of separately checking the "recurse" and "files_from" vars. This also simplifies the case where the code wants to send the implied dirs without recursing. Wayne Davison 2005-01-19 00:24:35 +00:00
  • 3a90ea0acd - Added new int, keep_dirs, which is set if either recurse or files_from was specified. - Pass the the sender the --no-implied-dirs option if --relative was specified (or implied) and the sender is not the client. Wayne Davison 2005-01-19 00:23:32 +00:00
  • 58af2f958c Fix typo in handling of lchown when the host operating system does not implement it. It should get mapped to chown. Paul Green 2005-01-18 21:19:42 +00:00
  • a8fd4161bb Mention the latest changes. Wayne Davison 2005-01-17 23:55:15 +00:00
  • 53f8519a38 For --inplace over protocol-version 29 or greater, read the type the type of basis file the generator is using from the socket and set the new updating_basis_file var. Wayne Davison 2005-01-17 23:11:26 +00:00
  • eb162f3b0a Mention the change in restrictions for --inplace. Also talk about how --backup makes an --inplace transfer more optimal. Wayne Davison 2005-01-17 23:11:22 +00:00
  • 4ce838e1f1 We don't need to complain about --inplace being combined with one of the --FOO-dest options anymore. Wayne Davison 2005-01-17 23:11:18 +00:00
  • a36ffd3910 Changed the checks for --inplace to use the new updating_basis_file variable. Wayne Davison 2005-01-17 23:11:13 +00:00
  • 17b5b32f75 For --inplace over protocol-version 29 or greater, tell the sender what kind of a basis file we've selected. Wayne Davison 2005-01-17 23:11:05 +00:00
  • 9b919d590a - Moved read_sum_head() to io.c (because the generator uses it too). - Tweaked the read_size var (formerly map_size) that gets passed to map_file(). Wayne Davison 2005-01-17 22:51:29 +00:00
  • 80264051d8 Tweaked the read_size var (formerly map_size) that gets passed to map_file(). Wayne Davison 2005-01-17 22:51:27 +00:00
  • 188fed9570 Moved read_sum_head() here from sender.c (because the generator uses it too) and improved it with better error checking. Wayne Davison 2005-01-17 22:51:24 +00:00
  • eae7165c79 Make sure that we don't compute a block size larger than MAX_BLOCK_SIZE. Wayne Davison 2005-01-17 22:51:21 +00:00
  • 54281fe733 - Changed the map_size parameter to map_file() to be an int32 named "read_size". - Changed some int vars in map_ptr() needed to be int32 vars. - All callers expect map_ptr() to return back the full "len" bytes, so we never shorten this value, even near EOF--the read will just get zero-padded, as needed. - Die with an error if someone calls map_ptr() with a len < 0, or if read_size is computed as a value < 0. Wayne Davison 2005-01-17 22:51:17 +00:00
  • 2b2ea368aa - Incremented the PROTOCOL_VERSION to 29. - Added define for MAX_BLOCK_SIZE. - Reorder the variables in the structure defs to sort by size. - Some vars in map_struct needed to be int32 instead of int. Wayne Davison 2005-01-17 22:51:15 +00:00
  • 67a28eb256 Got rid of an unneeded assert(). Wayne Davison 2005-01-17 22:25:53 +00:00
  • f5d96a6f80 Added a commented-out RSYNC initialization that uses valgrind. Wayne Davison 2005-01-17 22:24:08 +00:00
  • 44ac015598 Put double-quotes around the --rsync-path arg. Wayne Davison 2005-01-17 22:23:09 +00:00
  • 89a9c0545e Collapsed two sed invocations into one. Wayne Davison 2005-01-17 06:00:44 +00:00
  • 3b26bba0c4 Fixed an off-by-one comparison against MAX_BASIS_DIRS. Wayne Davison 2005-01-15 21:23:04 +00:00
  • 2be2fb3ed3 Use the new FNAMECMP_BASIS_DIR_HIGH to improve a compare-dest check. Wayne Davison 2005-01-15 21:17:39 +00:00
  • e341588a8a Define FNAMECMP_BASIS_DIR_LOW and FNAMECMP_BASIS_DIR_HIGH instead of FNAMECMP_BASIS_DIR. Wayne Davison 2005-01-15 21:17:36 +00:00
  • c56595d749 Make sure that the generator sent us a valid index into basis_dir[]. Wayne Davison 2005-01-15 21:14:27 +00:00
  • ce0b384fa6 Made basis_dir_cnt non-static. Wayne Davison 2005-01-15 21:12:58 +00:00
  • dfd7d541b0 Made compare_dest non-static. Wayne Davison 2005-01-15 20:54:25 +00:00
  • ded4daf049 Some inplace-checking code should have been inside the #ifdef that handles the case where --inplace cannot be supported. Wayne Davison 2005-01-15 20:22:54 +00:00
  • 361428213b Simplified the logic in allow_access() and tweaked some whitespace. Wayne Davison 2005-01-15 20:06:48 +00:00
  • 12a79db2f6 Simplified one "if" that checks if one of the --FOO-dest options was specified. Wayne Davison 2005-01-15 04:40:06 +00:00
  • e8b155a3be Improved the docs on --delete and --delete-after. Wayne Davison 2005-01-14 19:49:35 +00:00
  • 716b46c550 - We now remove the DEST~old~ dir instead of renaming it to DEST~new~. - Improved the usage message and added an introductory comment. Wayne Davison 2005-01-14 19:23:20 +00:00
  • 9d954dca8c Make sure that the presence of a partial-dir file is noted, even if --whole-file is in effect. Also needed to force statret to 0 when using a partial-dir file. Wayne Davison 2005-01-14 18:20:46 +00:00
  • 7162c65df7 We also need the "name pipe" from the generator to the receiver when --partial-dir was specified. Wayne Davison 2005-01-14 18:18:31 +00:00
  • bd9fca4708 For the "unexpected tag" or "multiplexing overflow" messages, we now output who_am_i() so we know who had the problem. Wayne Davison 2005-01-14 18:17:32 +00:00
  • 14d496cc8b Aid forward-compatibility in the include/exclude syntax by putting a "- " in front of any name that starts with a '+' or a '-' and would otherwise be unprefixed (the old code only did this if the '+' or '-' was followed by a space). Wayne Davison 2005-01-13 22:01:56 +00:00
  • 106a8ad918 - Allow multiple source paths to be specified (and checked). - Allow spaces and a few other extra chars in file names. - For safety, disallow any option that takes an arg. This should be improved in the future because it blocks options such as --block-size=N, but without this rule the user could specify something like --files-from=FILE or --backup-dir=DIR and have it affect files outside the desired SUBDIR restriction. - Switched to SSH_CONNECTION from the deprecated SSH_CLIENT. - Strip "::ffff:" from the start of an IP from SSH_CONNECTION. Wayne Davison 2005-01-12 19:20:07 +00:00
  • 44a82a175d This is Joe Smith's rrsync (restricted rsync) perl script. Wayne Davison 2005-01-12 18:48:26 +00:00
  • 99248631aa A simple Makefile. Wayne Davison 2005-01-11 18:37:37 +00:00
  • 2836ee9b02 A debug program to help diagnose data-transfer corruption problems. Wayne Davison 2005-01-11 18:36:31 +00:00
  • d82773ffe9 Fixed the file_checksum1() function that is compiled only when TEST_MDFOUR is defined: it did not have the fix that the main rsync code got back in protocol 27 to properly handle files that are a multiple of 64-bytes long. Wayne Davison 2005-01-10 20:52:08 +00:00
  • ec626b3f0e Two more fixes. Wayne Davison 2005-01-10 10:08:59 +00:00
  • 997d9ea67f Added a "overwriting_basis" arg to finish_transfer(). Wayne Davison 2005-01-10 10:03:12 +00:00
  • aec6b9f86f - Call finish_transfer() with its new arg. This ensures that we only create a backup file if we're about to overwrite the basis file. - Only clear make_backups for the redo phase if partial_dir isn't set. Wayne Davison 2005-01-10 10:03:10 +00:00
  • d45898df80 Call finish_transfer() with its new arg. Wayne Davison 2005-01-10 10:03:06 +00:00
  • b0da4b23a0 Moved the reading of the final MSG_DONE message here from main.c (this message comes from the receiver). This ensures that any redo-pass files and delete-after processing are known to be done prior to the start of our hard-link and dir-time-munging loops. Wayne Davison 2005-01-10 00:31:55 +00:00
  • bf18b7ca67 Moved the reading of the final MSG_DONE message from here to generator.c (the message comes from the receiver). This ensures that any redo-pass files and delete-after processing are known to be done prior to the start of the hard-link and dir-time-munging loops. Wayne Davison 2005-01-10 00:31:55 +00:00
  • cc17fbfe7d Mention new atomic-rsync script. Wayne Davison 2005-01-06 17:15:36 +00:00
  • fa170b2e5a A perl script to effect an atomic transfer of a set of files. Wayne Davison 2005-01-06 17:13:31 +00:00
  • c61ba345f2 Decided that a cross-compling configure should default the chown-follows-symlinks check to "yes". Wayne Davison 2005-01-03 23:01:19 +00:00
  • ae09fb1f91 Improved the last addition. Wayne Davison 2005-01-03 22:49:14 +00:00
  • 70a2c84cd2 If lchown() is not available, we only skip the call to chown() for a symlink if it will try to follow the symlink (as a normal OS should). Wayne Davison 2005-01-03 22:47:55 +00:00
  • 21524e3083 - Added a test to see if chown() follows symlinks. - Check for getpgrp() and tcgetpgrp(). Wayne Davison 2005-01-03 22:42:55 +00:00
  • fbe2aba2e7 Mention lchown() fix. Wayne Davison 2005-01-03 21:07:02 +00:00
  • a41a1e8718 If there is no lchown(), don't try to set the user & group of a symlink. Wayne Davison 2005-01-03 21:05:39 +00:00
  • 051547603a Moved kluged (conditional) define of lchown from rsync.h to syscall.c. Wayne Davison 2005-01-03 21:03:33 +00:00
  • 80a25bb880 Backed out changes to send_deflated_token() that surrounded the call to deflate(..., Z_INSERT_ONLY) -- the underlying bug was caused by the zlib code not handling Z_INSERT_ONLY in the case where the server has disabled compression. Wayne Davison 2005-01-02 09:08:59 +00:00