Peter Keane wrote:
> The trash feed is primarily for recovering a soft-deleted item correct?
> The OS analogy is a bit strained since the OS "trash" is the means by
> which you get rid of stuff and only secondarily provides a means to
> retrieve things.  Something like "removed" or "archived" or some such
> might be more descriptive.
> 

Yeah, I'll be working on the descriptive test around that a bit.  It's
not just for soft-delete.  The entry could have been permanently removed
from the underlying data store and added to a statically generated trash
file, primarily for the purpose of keeping it around so it could be
recreated later as a new entry.

> This "trash/removed/archived" feed is optional, correct?  I.e., it'll be
> used by server implementations optimized for soft-deletes (and wouldn't
> likely be used for comment spam, for instance)?
> 

Right. The trash feed is primarily intended to be used by the server to
provide the client a means of recovering deleted entries.  For things
like comment spam, the trash feed is not going to be needed.

> Also, since deleted-entry has an atom:id I can happily keep thinking of
> it as a kind of entry (quacks like a duck... ;-)), so I'll rescind my
> previous compliants.

:-). quacks like a duck with significantly fewer feathers.

> 
> All-in-all this looks like it covers the cases described in OAI-PMH
> (which from my POV is a very good thing).
> 

Good to hear. FWIW, after reading through the links you provided, I had
come to the same conclusion.

- James

> thanks!
> peter keane
> 
> On Wed, 2 Jan 2008, James M Snell wrote:
> 
>>
>> +1. I was already thinking that some text was needed describing the
>> relationship between these.
>>
>> - James
>>
>> Mark Nottingham wrote:
>>> Hmm. It seems like you need to say something to this; otherwise people
>>> may get the feeling they can use them separately.
>>>
>>> Cheers,
>>>
>>>
>>> On 03/01/2008, at 3:52 PM, James M Snell wrote:
>>>
>>>>
>>>> Mark Nottingham wrote:
>>>>> This looks good, except I'm having trouble with the trash feed; is a
>>>>> client who wishes to implement this spec required to look both for
>>>>> deleted-entry and link/@rel="trash", polling that if present? If
>>>>> so, it
>>>>> seems overly cumbersome to implement two mechanisms.
>>>>>
>>>>
>>>> The client can look for either.  Essentially, the deleted-entry element
>>>> tells you only that the entry was removed from this particular
>>>> feed.  It
>>>> does not tell you if the entry exists in some other location.  The
>>>> trash
>>>> feed is orthogonal to deleted-entry and provides a location where
>>>> things
>>>> like soft-deleted items can be collected.  For instance, suppose we
>>>> have
>>>> an Atompub collection managed by multiple authors, Bob and Joe.  Bob
>>>> deletes an entry. Joe discovers that that the entry has been deleted
>>>> when he sees the deleted-entry element.  If Joe wishes to recover the
>>>> deleted entry, he can go to the trash feed and look for it there.  A
>>>> client could poll the trash feed to discover deleted items if it wishes
>>>> to do so, but looking for the deleted-entry items would likely be more
>>>> efficient.
>>>>
>>>> - James
>>>>
>>>>> BTW, "trash" is culturally specific; how about "garbage"? <1/2 wink>
>>>>>
>>>>> Cheers,
>>>>>
>>>>>
>>>>> On 03/01/2008, at 1:15 PM, James M Snell wrote:
>>>>>
>>>>>>
>>>>>> Reviving the Tombstones draft. Consider this a work in progress.
>>>>>>
>>>>>> http://www.ietf.org/internet-drafts/draft-snell-atompub-tombstones-03.txt
>>>>>>
>>>>>>
>>>>>>
>>>>>> There are two significant changes:
>>>>>>
>>>>>> 1. deleted-entry is now part of the Atom namespace
>>>>>> 2. a "trash" link relation is registered to point
>>>>>>   to trash feeds.
>>>>>>
>>>>>> The trash feed approach and deleted-entry element approaches are
>>>>>> definitely compatible with one another.  They can be used together or
>>>>>> independently.
>>>>>>
>>>>>> - James
>>>>>>
>>>>>
>>>>>
>>>>> -- 
>>>>> Mark Nottingham     http://www.mnot.net/
>>>>>
>>>>>
>>>
>>>
>>> -- 
>>> Mark Nottingham     http://www.mnot.net/
>>>
>>>
>>
> 

Reply via email to