Matt Benjamin wrote:
Based on comments made in the previous round of discussion, which seems
now to have died down, I'd like to suggest an interface for further
discussion.

The intention is not to innovate beyond the ideas for which there
appeared to be consensus (supporting snippet included).

Thanks to Tom Kaiser, for initial check-over and comments.

Thanks,

Matt
The devil is in the details. While the XG might be beneficial to help you solidify your thoughts, I recommend that you focus on the English description of each message exchange and how it will be used.
I do have two technical comments:

  1. I think the file server should be able to cancel the callback
     contract as part of the delivery of any event.
  2. The AFSCB_Event_XXX messages that you list parallel the RPCs made
     to the file server by clients.  However, they do not necessarily
     describe the notifications that need to be sent to the receiver of
     the callback.   An RXAFS_RemoveFile RPC effects two objects, the
directory that contains the entry and the file that was removed. At present there are two callbacks that are sent. One for each object. I think there needs to be two different messages there. One describing the change to the directory object and the other
     describing the removal of the file.  Note that removal of the file
     is going to be a message that cancels the callback contract since
     the file will no longer exist.

Spending the time to produce an Internet Draft style document will make it a lot easier to determine what has been missed. The XG can be an appendix to that document but it cannot be a replacement for it.

Jeffrey Altman


Attachment: smime.p7s
Description: S/MIME Cryptographic Signature

_______________________________________________
AFS3-standardization mailing list
[email protected]
http://michigan-openafs-lists.central.org/mailman/listinfo/afs3-standardization

Reply via email to