cannot pickle universe object

[GoogleCodeExporter]

What steps will reproduce the problem?
1. Create a universe object by passing a psf file and a dcd file
          u = universe(psf, dcd)
2. Try to serialize this object by using python pickle module
          u_blob = pickle.dumps(u)

What is the expected output? What do you see instead?
Expected output should be a byte stream, but I got error message 'TypeError: 
'AtomGroup' object is not callable'

What version of the product are you using? On what operating system?

MDAnalysis 0.8.0
Python 2.7.6 / Python 2.6.2
Arch Linux / CentOS

Please provide any additional information below.

Original issue reported on code.google.com by charlesz...@gmail.com on 27 Mar 2014 at 9:18

[orbeckst]

Is this going to be solved by #643 ?

[richardjgowers]

Sort of, Universe pickle will just call topology.pickle and
trajectory.pickle. 643 is most of the hard work though

On Sat, 30 Apr 2016 00:35 Oliver Beckstein, notifications@github.com
wrote:

Is this going to be solved by #643
#643 ?

—
You are receiving this because you are subscribed to this thread.
Reply to this email directly or view it on GitHub
#173 (comment)

[dotsdl]

As currently written Topology can be pickled, but the 10M atom system I've been playing with takes much longer to deserialize from disk and the filesize of the pickle is nearly 7 times larger than the JSON. I think if we decide to stick with JSON as the preferred serialization for topologies, we can just define the JSONified topology as the output of __setstate__ (and likewise de-JSONifying for __getstate__).

Trajectory pickling is probably just going to have to be pickling/unpickling the filepaths. Will be sufficient for multiprocessing or distributed computing on a common filesystem.

[kain88-de]

@dotsdl did you enable compression for the pickle? The standard pickle protocol is super inefficient. If you change it the results are comparable to JSON. I had the same problems deciding how to store offsets in the xdrlib refactor.