rsvndump [options] url
rsvndump dumps a Subversion repository without having actual access to the repository data, as required by the "svnadmin(1) dump" command.
If invoked with valid options, rsvndump will print a dump file to stdout and optional progress to stderr.
Since rsvndump is also able to dump subdirectories of a repository, the revision numbers in the dump don’t necessarily reflect the original revision numbers. For more information on this, please refer to the Revision Numbers section.
There are also some additional differences to a "normal" dump generated by svnadmin(1), which are listed in Differences to Svnadmin Dump.
Most of the options can also be found in svnadmin(1) and svn(1) and should be semantically equivalent, unless not stated otherwise.
- -h, --help
Print a nice help screen
- -q, --quiet
Don’t print any progress
- -v, --verbose
Print detailed progress, similar to "svnadmin(1) load". Can be specified multiple times.
- -u, --username username
User name for repository authentication.
- -p, --password password
Password for repository authentication.
- -r, --revision X or X:Y
Specify the revision (or revision range) that should be dumped. X and Y can either be a number or "HEAD". The default revision range is "0:HEAD", i.e. the complete revision history.
In contrast to svnadmin(1), dumps that were generated using this option can always be loaded into a new repository, (unless --incremental is given, of course). However, the resulting dump may not exactly represent the original history. Please take a look at Differences to Svnadmin Dump for further information.
Use text deltas instead of full texts in dump output
Create incremental output, suitable for concatenation. This results in a dump that does not contain a dumpfile header and no full base revision if the revision range does not start at 0.
Don’t cache the authentication tokens provided by the user, e.g. user name and password, or manual SSL certificate validation. This is equal to the corresponding option in svn(1).
Instead of prompting, the program will exit with an error. This is equal to the corresponding option in svn(1).
- --config-dir dir
Use the given Subversion configuration directory when opening a session to the repository. This is equal to the corresponding option in svn(1).
- --prefix prefix
Prepend a given prefix to every node which is dumped. The first revision dumped will also contain the paths necessary to create the prefix.
Please note that is assumed that prefix is really just a prefix string, with directories separated by "/". Therefore you need to append a "/" at the end of the string to make it a directory. For example, a prefix of myrepo/old_ might result in the following layout:
myrepo |- old_branches |- old_tags |- old_trunk
Keep the revision numbers in the output in sync with the repository. This is done by inserting empty revisions for padding if necessary.
Don’t print the dumpfile header if --incrementals is given and the revision range is not "0:X". This is useful if you really want to append an incremental dumps to an existing file.
- -n, --dry-run
Don’t fetch text deltas, resulting in a dump without file contents. This is mainly used for debugging purposes, as the program operates normally but runs much faster.
Replaces all file and directory names with random strings. This is useful for bug reports in combination with --dry-run.
The revision numbers in the dump output depend on the options and the path that are given to rsvndump. If you are dumping the root of a repository, you don’t need to worry about revision numbers out of sync, of course. If you are dumping a subdirectory, only the revisions that changed this subdirectory will occur in the dump output. The revision numbers in the dump are strictly sequential, so they will differ from the original ones.
If you need the keep the revision numbers from the original repository (e.g, if a bug tracker depends on them), you can use the --keep-revnums flag. It pads revisions that did not change the subdirectory with empty revisions. They don’t have an author or date property, but contain the log message "This is an empty revision for padding.".
Differences to Svnadmin Dump
The output generated by rsvndump may differ from the one generated by svnadmin(1) because rsvndump may handle file or directory copies different than svnadmin(1). The latter does not support dumping of subdirectories within a repository out of the box. Instead, the svndumpfilter(1) tool will do this job. However, sometimes subdirectories cannot be filtered exclusively with svndumpfilter(1) because they have been copied from another place.
Since rsvndump has been designed to allow dumps of subdirectories even if your repository access is limited to this subdirectory (in fact, that was the primary reason this project exists), it will not access other subdirectories than the one you want to dump. Example given, dumping a branch will most likely not result in the full history of the branch as it might have been copied from trunk at some point in time.
Thus, rsvndump will generally replace a copy action by a simple add operation if both of the following conditions are true:
- The source of the copy is outside the directory tree which is being dumped
- The source of the copy is not included in the dump because the revision range has been limited using the --revision flag
The second condition is without effect if the --incremental flag is given, so that incremental dumps yield the same result as normal dumps.
Maybe I should also note that the dump will only contain the repository’s UUID if the following conditions are true:
- The root of the repository is being dumped
- There is no user prefix
If TMPDIR is set, it specifies the directory to use for temporary files. Otherwise, /tmp is used.
0 on success, 1 on failure. Any error messages will be printed to stderr.
svn(1), svnadmin(1), svndumpfilter(1)
Written by Jonas Gehring, <email@example.com>
Copyright © 2008-2011 Jonas Gehring <firstname.lastname@example.org>. Released under the GNU General Public License.