Coda File System

Re: Strange timeouts / no such device

From: Greg Troxel <gdt_at_ir.bbn.com>
Date: Fri, 23 Mar 2007 18:28:29 -0400
"Stephen J. Turnbull" <stephen_at_xemacs.org> writes:

> I see no way to distinguish "rm --please-make-the-conflict-go-away"
> from "rm --please-make-the-file-go-away".  Remove all versions?

Sorry, I was unclear.  In a directory, one has a file foo.  It's
perhaps 10k of text.  Now, you get a local-global conflict, say
because you edited it disconnected and someone else wrote it.
Now, it's a dangling symlink.

You do 'cfs beginrepair foo', or something like that.

Now you have a *directory* foo, and files
  foo/local
  foo/global

If you "rm foo/local", that's like cfs discardlocal.
"cfs endrepair foo" would note that there's no longer a conflict and
you'd have a file foo gain, with global contents.

> I think a better UI would be to merge the two files (a la merge(1))
> into file/result or something like that.  I'm not sure how that would
> deal with the problem above, though.

That's sort of ASR, which is ok if you know that's what you want.

> Do Unix tools grok the semantics of Schroedinger's cat?  I suspect
> that's what we need.  It would be very nice to be wrong about that. ;-)

Well, coda actually feels a bit like unevaluated quantum state.  You
think your files are there, and that you'll be able to get at them
when disconnected, but then you see what happens when the time comes.
The odds seem much better than they used to be, though.


Received on 2007-03-23 18:29:53