Added Princeton copyright to sfa files, and modified
src/sfa/README.txt to include the addition of the copyrights as a
modification from the original source. This header should be added
every time.

Added BBN copyright to BBN written files that did not have the
copyright header.

getslicecred prints the slice credential for debugging.

Added to base class, reference implementation and sample printing
AM. Also added logger.debug statements to print the omni client
exception instead of hiding it.

All omni methods that construct URNs now use
framework.slice_name_to_urn instead of namespace.long_urn. The default
implementation in Framework_Base uses namespace.long_urn. framework_pg
overrides this behavior to create a URN based on the domain name of
the slice authority.

Additionally added some debug printouts for an exception to aid in
debugging. Added pretty printing of status dict for easier reading.

Revealed in testing, the requirement for Python 2.6 was not
documented. Added the requirement to README.txt. Also copied the
version testing block from gam.py into init-ca.py, gch.py, and
client.py.

get_slice_cred and create_slice are implemented based on the pattern
in the ProtoGENI test programs (emulab/protogeni/test). See
createsliver.py and test-common.py in the emulab source tree.

ProtoGENI RSpec requests are hard to generate from the advertisement
in a meaningful way. This is made even harder with the omnispec
conversion. Punting the problem by hardcoding a simple request RSpec
that will allow testing for now.

URNs may be site dependent, so the framework should be responsible for
creating the URN from a name. In the case of ProtoGENI, the URN
depends on the slice authority, and the user should not need to know
the syntax of a slice URN. The base class implementation just converts
directly to URN via namespace.long_urn. In ProtoGENI the domain of the
slice authority is computed based on the slice authority URL, and the
URN is constructed accordingly.

Elementtree makes it hard because of the lost xmlns information. Will
experiment with other alternatives to see if something else will work better.

Skeletal implementation to read pg rspecs into omnispec is
complete. Information is lost, however. Translation back to pg rpsec
has not been implemented.

GetVersion returns the pg style code/value/output dict on pgeni3
because it is running older code. To be removed once pgeni3 is
upgraded.

Gets the list of component managers from the ProtoGENI
clearinghouse. Figures out which are AMs. Returns only those that are
AMs.

If timeout is passed, use the value as the timeout for the underlying
socket. This will avoid waiting for xml rpc servers that aren't really
there or are misbehaving.

The ProtoGENI CMs are retrieved from the ProtoGENI CH. Need to filter
this list down to those sites that have an AM running.

No other backend methods work yet. Most will result in errors.

By default logging is at info level. If --debug command line switch is
used logging is at debug level. The logger named "omni" is configured
so all subordinate logging should use "omni." as the prefix to get the
proper logging hierarchy.