Subversion 1.7 arrives with improved metadata storage

The Apache Subversion developers have announced the release of version 1.7 of the source code control system. Among the new features of version 1.7 is a reworking of how Subversion handles metadata for working copies. Instead of the current scheme of a ".svn" file per directory of a working copy, there is now a single ".svn" directory with an SQLite database for all the metadata. Third-party tools could work with this data, but the developers say the potential for data corruption is high and do not recommend it. They also note that backups and duplicates of the working copy should only be taken when it is not being accessed by a Subversion process. The new metadata management should improve performance, but more importantly, provides the basis for future new features.

Other new features include improved HTTP protocol usage between the client and server which reduces the number of roundtrips needed to establish connections, a new remote dump tool and a new "svn patch" command which can apply unidiff format patches to a working copy. Enhancements include the addition to svn log's of the ability to print diffs, svn update being able to create patent directories, a new svn relocate command, a smarter svn switch which checks ancestry and a filesize subcommand for svnlook which returns size information on paths in the repository.

Full details of these and other changes in Subversion 1.7 can be found in the release notes. Subversion 1.7 has now moved to being licensed under the standard Apache Licence version 2 and its collective copyright is owned by the Apache Software Foundation. The developers have also noted that Subversion 1.5 is no longer supported with the release of 1.7, and they will not be making any 1.5.x bugfix releases from now on.

Source code for Subversion 1.7 can be downloaded from the project's source download page. Unofficial binary versions are also listed on the site and are expected to begin updating to 1.7, starting in a week or so.

(djwm)