Combining 2 responses...

===
To clarify, if you receive the same item from the same person in 2 different shares, 1 RW, the other RO, the RW privilege on the item should win.

On Mar 16, 2007, at 10:33 AM, Brian Moseley wrote:

On 3/15/07, Mimi Yin <[EMAIL PROTECTED]> wrote:

Morgen, I think that's the desired behavior? If you have 1 item in 2
collections with conflicting privileges, we want the more liberal
privilege to win on that item, namely read-write.

completely disagree.

===
With respect to the below, I think we're getting into the realm of theory and best practices. Practically speaking, what is wrong with no-item-soup' on Chandler Server for Preview? Ted's response


To clarify, what's 'not-OK' about 'no-item-soup' on Chandler Server?

an item is a fundamental piece of data with its own identity, and if i
share it with you, you should see whatever updates i make to it and
vice versa, without either the client or the server having to
synchronize my copy with your copy or any other funny business.

===
Morgen, Sheila and I have discussed Contact-Sharing scenarios where a per-account item soup is precisely what you want / expect as an user. (Namely, sharing contacts via item references. I share an item with you which references a contact item you and I both have.)

Or to be less wedded to a particular solution and more goal-oriented, even when two items have the same UUID, there are times when you want to distinguish between two items, precisely by who 'owns' which version of the item. I don't know that a per-account item soup is the answer to that, but the statement below is very broad and I'm not sure how it plays out when taking specific scenarios into consideration.

On Mar 16, 2007, at 11:05 AM, Brian Moseley wrote:
On 3/16/07, Ted Leung <[EMAIL PROTECTED]> wrote:

Can you elaborate on you the reasoning here?  I'm not challenging it,
but I want to understand what the issues are.

i think it artificially limits our capabilities. in a world of shared
data, ownership is not a very powerful or useful concept. bucketing
data in that way just introduces artificial constraints and makes the
code for implementing sharing features more complex.

Mimi
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _

Open Source Applications Foundation "Design" mailing list
http://lists.osafoundation.org/mailman/listinfo/design

Reply via email to