option.

XXX This will probably go away once we decide how we really want to do
this.

storages, it's much easier to specify those in a config file.

XXX This still needs a unit test suite.

records (requiring occasional seeks).

  severity level.  This allows the common names to be used for
  severities, not just integers.  This can make configuration data
  less obscure.

- Refactor initialize():
  get_environment_info() pulls information from the environment
  variables, and this gets passed to initialize_log(), which does
  nothing to determine the source of the configuration data.

- New method initialize_with_config():  Load information not provided
  by the environment from a ZConfig section.  This allows the
  environment to override a config file, and avoids having the
  application deal with the specific information needed to configure
  logging.

port) tuple.

  rilog()) into methods, so the log message can include the storage
  name.

- Change the trace filename to include a hyphen and the storage name.

This addresses ZODB SF bug 641925.

ConnectThread from hanging the application on exit.

Add -S to usage message.

Rephrase error message about address in use.

zdaemon/zdaemon.py.

Also get rid of the sys.path manipulation; it doesn't work (since the
C extensions are not there).

You can now do this with the -u USER zdaemon/zdaemon.py.

that don't use it -- gcc complains if you do.

seems to work.

XXX Not sure if berkeley still works.

Put a MinPO in the root, so that we don't actually modify the root.
Print the elapsed time for zeoup to run.

Indentation of the 2nd line is a lose, because the 2nd line can be
long.  Instead put a blank line after a record.

Add fstools module with simple utilities for jumping around in a Data.fs.

storage and creates a new one.  We need this here so Berkeley storage
tests can do something different.

zodb3-release branch.

ZODB3-restore-debug branch, and genericized, which checks for proper
backpointers when recovering undos w/ commit and abort versions.