Hey guys, VFS has languished in a state of neglect for the last 2 years. What is currently in the repo is the output of 2 and a bit sprints - and some patches from canonical so it works for their immediate needs for launch pad. In retrospect - we got a bit excited that first sprint in Hobart, and we probably shouldn't have pushed what we had into the main twisted repo until I could show the commitment to get it into a more solid form. VFS trunk, currently defines a synchronous API. The API isn't complete. The filesystem backend uses its own utils, instead of the established twisted.python.filepath. Variable naming is all over the place, and the code quality is pretty poor - coming from a sprint were the focus was on how much could we get working before the weekend is out. The branch to rectify these issues is: branches/makeVFSAsync-1223-2 Things done so far in the branch: * The API is better defined and is async. There's still some debate over the API, particularly over meta-data. But I think its a good step forward and if merged the interface could be upgraded from obscenely unstable to unstable. * The memory and filesystem backends have been ported to meet the new API. * The filesystem implementation uses twisted.python.filepath for most operations. Things still left to do for the branch to be merged: * clean up the adapter tests: http://twistedmatrix.com/trac/ticket/1264 * port the existing adapters to use the new async backends * clean up variable/module naming to be more consistent * remove all of the now unused code * ensure all patches since branching from canonical are incorporated cheers, Andy.