Question about activity:target

[James Walker]

After a few re-readings on activity:target [1] , we were wondering if
we could get some clarification. Would activity:target be appropriate
for something like "mentions" or "replies" ? The section claims that
"The activity:target extension element MUST NOT be used for indirect
objects that are not targets. " - and the referenced example uses a
"photo album" as an example target.

However, before that it says "but its meaning is somewhat similar to
the English preposition "to"".

The usage we are considering would be something like the following:

James (actor) has posted (verb) a note (object) "to" Evan (target)

Is this a fair interpretation? Also, would it be acceptable to have
multiple targets? (i.e. mention multiple users in a post).

Thanks :)

[1] http://martin.atkins.me.uk/specs/activitystreams/atomactivity#anchor7

--
James Walker :: http://walkah.net/

[Martin Atkins]

On 03/03/2010 11:26 PM, James Walker wrote:
> After a few re-readings on activity:target [1] , we were wondering if
> we could get some clarification. Would activity:target be appropriate
> for something like "mentions" or "replies" ? The section claims that
> "The activity:target extension element MUST NOT be used for indirect
> objects that are not targets. " - and the referenced example uses a
> "photo album" as an example target.
>
> However, before that it says "but its meaning is somewhat similar to
> the English preposition "to"".
>
> The usage we are considering would be something like the following:
>
> James (actor) has posted (verb) a note (object) "to" Evan (target)
>
> Is this a fair interpretation? Also, would it be acceptable to have
> multiple targets? (i.e. mention multiple users in a post).
>

When used with the verb "to post" we generally assume target to be the
thing that the object was posted in to, such as a blog or a photo album.

It seems like what you're trying to model here is something like a
Twitter @reply, which is not necessarily in response to a particular
message but is rather just addressed to a specific person.

This does seem to be a hole in the spec right now, but given the
established meaning of "target" when used in conjunction with the "post"
verb, I think it would be better to attack this in a similar way to how
we attack traditional comments -- by adding a property to the note
itself which says who it is addressed to.

For a traditional comment this is done with the Atom Threading
Extensions in-reply-to element, but that element assumes that the
comment is in reply to some other object, rather than just being a
message addressed to a specific person. I think to represent this we'd
need to make something up, since I don't know of any existing Atom
extension for representing that an entry is addressed to a particular
person.

[James Walker]

Thanks for the feedback, Martin. This is more inline with our original
interpretation.

For context, yes we at status.net are using ActivityStreams (with both
PuSH and Salmon) for federated status updates. We're already using
atom threading for the case where we know for sure that it's
in-reply-to a specific notice. However, for the more casual "mention"
we'd like to be able to verify that the post was actually intended
(from the source's perspective) for the receiving account.

We'd already introduced our own element ( <ostatus:attention /> ) for
this purpose - since threading doesn't work and it didn't seem like
activity:target wasn't quite right. Sounds like that was the right
approach. We revisited it because we'd rather not have our own
elements... so if this is something that would have general use for
activitystrea.ms we'd be all for adopting it.

Anyone else come across a need / use case for "directed activity" ?

[Martin Atkins]

On 03/04/2010 09:37 AM, James Walker wrote:
>
> Thanks for the feedback, Martin. This is more inline with our original
> interpretation.
>
> For context, yes we at status.net are using ActivityStreams (with both
> PuSH and Salmon) for federated status updates. We're already using
> atom threading for the case where we know for sure that it's
> in-reply-to a specific notice. However, for the more casual "mention"
> we'd like to be able to verify that the post was actually intended
> (from the source's perspective) for the receiving account.
>
> We'd already introduced our own element (<ostatus:attention /> ) for
> this purpose - since threading doesn't work and it didn't seem like
> activity:target wasn't quite right. Sounds like that was the right
> approach. We revisited it because we'd rather not have our own
> elements... so if this is something that would have general use for
> activitystrea.ms we'd be all for adopting it.
>
> Anyone else come across a need / use case for "directed activity" ?
>

It sounds like your "attention" element would be generally useful as a
standalone Atom extension, like Atom Threading Extensions is.

Would you be interested in writing a draft spec for that element? The
spec document could probably be similar in structure to the Atom
Threading Extensions document.

[John Panzer]

Bikeshedding:  I think it'd be better to have something equivalent to ostatus:attention as a link with a new rel (the relationship being, the target href is mentioned in this entry).  

As an aside, per a discussion I just had with Will Norris, and want to capture before I forget:

A good reason why this has to be in the payload for purposes of attention pings, as opposed to just assuming that the recipient is using unique URIs for each salmon-mention endpoint:  You want this to be covered by the Salmon signature mechanism, which only covers the payload data.  If you don't sign it, you're vulnerable to an attack in which someone damages your reputation by copying your salmon and pretending that you're sending them to random endpoints who aren't really mentioned in the text.

--
You received this message because you are subscribed to the Google Groups "Activity Streams" group.
To post to this group, send email to activity...@googlegroups.com.
To unsubscribe from this group, send email to activity-strea...@googlegroups.com.
For more options, visit this group at http://groups.google.com/group/activity-streams?hl=en.

[James Walker]

On Thu, Mar 4, 2010 at 4:32 PM, John Panzer <jpa...@google.com> wrote:
> Bikeshedding:  I think it'd be better to have something equivalent to
> ostatus:attention as a link with a new rel (the relationship being, the
> target href is mentioned in this entry).

Hrm. I'm not opposed to this switch - but can you expand as to why?
(off list even, if it's not appropriate to AS specifically).

> As an aside, per a discussion I just had with Will Norris, and want to
> capture before I forget:
> A good reason why this has to be in the payload for purposes of attention
> pings, as opposed to just assuming that the recipient is using unique URIs
> for each salmon-mention endpoint:  You want this to be covered by the Salmon
> signature mechanism, which only covers the payload data.  If you don't sign
> it, you're vulnerable to an attack in which someone damages your reputation
> by copying your salmon and pretending that you're sending them to random
> endpoints who aren't really mentioned in the text.

Yes, exactly... we check (and will drop) things that his our salmon
endpoint that don't either a) contain an atom threading in-reply-to or
b) an <ostatus:attention /> that matches (and we can have several
attentions).

So glad that you guys agree! (John: salmon-wise this is part of why i
don't think mentions / replies need separate endpoints - we've
combined them in our implementation).

--
James Walker :: http://walkah.net/

[Will Norris]

On Thu, Mar 4, 2010 at 1:45 PM, James Walker <wal...@walkah.net> wrote:

On Thu, Mar 4, 2010 at 4:32 PM, John Panzer <jpa...@google.com> wrote:
> Bikeshedding:  I think it'd be better to have something equivalent to
> ostatus:attention as a link with a new rel (the relationship being, the
> target href is mentioned in this entry).

Hrm. I'm not opposed to this switch - but can you expand as to why?
(off list even, if it's not appropriate to AS specifically).

 If for no other reason than it prevents having to mint a new element in yet another namespace.  Is there any reason why a simple link would be insufficient?  Is there a need to provide additional metadata about the mentioned party, aside from a URL?

-will

[John Panzer]

On Thu, Mar 4, 2010 at 1:45 PM, James Walker <wal...@walkah.net> wrote:

On Thu, Mar 4, 2010 at 4:32 PM, John Panzer <jpa...@google.com> wrote:
> Bikeshedding:  I think it'd be better to have something equivalent to
> ostatus:attention as a link with a new rel (the relationship being, the
> target href is mentioned in this entry).

Hrm. I'm not opposed to this switch - but can you expand as to why?
(off list even, if it's not appropriate to AS specifically).

Links are more constrained and there's already well established paths through (most) libraries that you can expect a collection of typed links to be easily accessible.  It's not a huge deal, obviously, but it's nice to have.  It also seems to me that an additional link relation is easier to mint than a new element.

[Chris Messina]

I added a page to the wiki that needs to be fleshed out:

https://activitystreams.pbworks.com/Privacy-and-Audience

Let's document these issues here and attempt to come to some kind of solution... whether it's link-based or requires a new element.

Chris

Chris Messina
Open Web Advocate, Google

Personal: http://factoryjoe.com
Follow me on Buzz: http://buzz.google.com/chrismessina
...or Twitter: http://twitter.com/chrismessina

This email is:   [ ] shareable    [X] ask first   [ ] private

[tyler gillies]

how do you insert

This email is:   [ ] shareable    [X] ask first   [ ] private

Everyone Loves Tea
http://www.everyonelovestea.com