Cedar-backup3

* Cleanup some maintainer-only (non-distributed) Makefile rules.
* Make changes to standardize file headers with other Cedar Solutions code.
* Add debug statements to filesystem code (huge increase in debug log size).
* Standardize some config variable names ("parentNode" instead of "parent").
* Fix util/test.py to return proper (non-zero) return status upon failure.
* No longer attempt to change ownership of files when not running as root.
* Remove regression test for bug 25 (testAddFile_036) 'cause it's not portable.
* Modify use of user/password in MySQL extension (suggested by Matthias Urlichs).
- Make user and password values optional in Cedar Backup configuration
- Add a few regression tests to make sure configuration changes work
- Add warning when user or password value(s) are visible in process listing
- Document use of /root/.my.cnf or ~/.my.cnf in source code and user manual
- Rework discussion of command line, file permissions, etc. in user manual
* Optimize incremental backup, and hopefully speed it up a bit (closes: 29).
- Change BackupFileList.removeUnchanged() to accept a captureDigest flag
- This avoids need to call both generateDigestMap() and removeUnchanged()
- Note that interface to removeUnchanged was modified, but not broken
* Add support for pre- and post-action command hooks (closes: 27).
- Added <pre_action_hook> and <post_action_hook> sections within <options>
- Updated user manual documentation for options configuration section
- Create new config.PreActionHook and PostActionHook classes to hold hooks
- Added new hooks list field on config.OptionsConfig class
- Update ActionSet and ActionItem in cli to handle and execute hooks
* Rework and abstract XML functionality, plus remove dependency on PyXML.
- Refactor general XML utility code out of config.py into xmlutil.py
- Create new isElement() function to eliminate need for Node references
- Create new createInputDom(), createOutputDom() and serializeDom() functions
- Use minidom XML parser rather than PyExpat.reader (much faster)
- Hack together xmlutil.Serializer based on xml.dom.ext.PrettyPrint
- Remove references to PyXML in manual's depends.xml and install.xml files
- Add notes about PyXML code sourced from Fourthought, Inc. in CREDITS
- Rework mysql and subversion unit tests in terms of new functions

* Fix broken call to node.hasChildNodes (no parens) in config.py.
* Make "pre-existing collect indicator" error more obvious (closes: 26).
* Avoid failures for UTF-8 filenames on certain filesystems (closes: 25).
* Fix FilesystemList to encode excludeList items, preventing UTF-8 failures.

* Remove bogus check for remote collect directory on master (closes: 18).
* Fix testEncodePath_009 test failure on UTF-8 filesystems (closes: 19).
* Fixed several unit tests related to the CollectConfig class (all typos).
* Fix filesystem and action code to properly handle path "/" (closes: 24).
* Add extension configuration to cback.conf.sample, to clarify things.
* Place starting and ending revision numbers into Subversion dump filenames.
* Implement resolver mechanism to support paths to commands (closes: 22).
- Added <override> section within <options> configuration
- Create new config.CommandOverride class to hold overrides
- Added new overrides field on config.OptionsConfig class
- Create util.PathResolverSingleton class to encapsulate mappings
- Create util.resolveCommand convenience function for code to call
- Create and call new _setupPathResolver() function in cli code
- Change all _CMD constants to _COMMAND, for consistency
* Change Subversion extension to support "fsfs" repositories (closes: 20).
- Accept "FSFS" repository <type> in <subversion> configuration section
- Create new FSFSRepository class to represent an FSFS repository
- Refactor internal code common to both BDB and FSFS repositories
- Add and rework test cases to provide coverage of FSFSRepository
* Port to Darwin (Mac OS X) and ensure that all regression tests pass.
- Don't run testAddDirContents_072() for Darwin (tarball's invalid there)
- Write new ISO mount testing methods in terms of Apple's "hdiutil" utility
- Accept Darwin-style SCSI writer devices, i.e. "IOCompactDiscServices"
- Tweak existing SCSI id pattern to allow spaces in a few other places
- Add new regression tests for validateScsiId() utility function
- Add code warnings and documentation in manual and in doc/osx
* Update, clean up and extend Cedar Backup User Manual (closes: 21).
- Work through document and copy-edit it now that it's matured
- Add documentation for new options and subversion config items
- Exorcise references to Linux which assumed it was "the" platform
- Add platform-specific notes for non-Linux platforms (darwin, BSDs)
- Clarify purpose of the 'collect' action on the master
- Clarify how actions (i.e. 'store') are optional
- Clarify that 'all' does not execute extensions
- Add an appendix on restoring backups

* Update docs to modify use of "secure" (suggested by Lars Wirzenius).
* Removed "Not an Official Debian Package" section in software manual.
* Reworked Debian install procedure in manual to reference official packages.
* Fix manual's build process to create files with mode 664 rather than 755.
* Deal better with date boundaries on the store operation (closes: 17).
- Add <warn_midnite> value in <store> configuration
- Add warnMidnite field to the StoreConfig object
- Add warning in store process for crossing midnite boundary
- Change store --full to have more consistent behavior
- Update manual to document changes related to this bug

* Fix boundaries log message again, properly this time.
* Fix a few other log messages that used "," rather than "%".

* Fix minor typos in user manual and source code documentation.
* Properly annotate code implemented based on Python 2.3 source.
* Add info within CREDITS about Python 2.3 and Docbook XSL licenses.
* Fix logging for boundaries values (can't print None[0], duh).