Re: Guaranteeing the Integrity of Bindings

"Slein, Judith A" wrote:

> Jim Amsden in his comments on the advanced collections spec said that he
> believes it is never possible to guarantee the integrity of bindings that
> cross servers.  So requiring servers to guarantee the integrity of any
> binding that they allow to be created will have the result that no one will
> ever implement cross-server bindings.
>
> The kind of case he had in mind (I think) was one where the server where the
> resource resides becomes inaccessible to the server where bindings resides.
> So at least while the 2 servers are out of touch, the binding will be
> broken.

Actually, it's more than that.  You're (apparently) presuming that the servers
can come up with some way of making DELETE on the resource trigger a DELETE on
the binding.  But suppose the resource is just a file, and it gets deleted from
outside the server? I am not aware of any OS where you can get notification
when a file is deleted.  In the single-server case, this is not a problem,
because the server can check to make sure the resource still exists before
doing anything to indicate that the binding exists.  In the cross-server
case...well, you could still do it, but it'd make things like PROPFIND
expensive (if a collection contains 15 cross-server bindings, you've got to do
15 HTTP requests to make sure they're all still there).

--
/================================================================\
|John Stracke    | http://www.ecal.com |My opinions are my own.  |
|Chief Scientist |===============================================|
|eCal Corp.      |I'm not a bibliophile, I'm a bibliophiliac. Put|
|francis@ecal.com|me in a bookstore, & my wallet bleeds.         |
\================================================================/

Received on Wednesday, 15 September 1999 15:36:40 UTC