Closed GoogleCodeExporter closed 9 years ago
Thanks for the comment!
Original comment by jpanzer@google.com
on 26 Feb 2010 at 1:47
That's actually closer to how we're handling things in OStatus / StatusNet ...
we
have a salmon link in each user's XRD ... and we also add an
<ostatus:attention>acct:user@example.com</ostatus:attention> for each user
mentioned
in an entry.
John can you clarify the reasoning behind multiple endpoints?
Original comment by wal...@gmail.com
on 26 Feb 2010 at 5:37
I don't have a strong feeling either way. I did want to clarify the two use
cases in the
document, because they have somewhat different requirements, but mainly it was
just that I couldn't prove to myself that there wasn't ambiguity. As long as
it would
not cause problems we could collapse these two.
Feedback that I've gotten is that people are worried about being able to
separate out
the non-conversational mentions (a lot of tweets are like this) from things
where the
user really intends to reply. But the metadata may be enough for this.
Thoughts?
Collapse these two endpoint relations into one relation? And if so, what
should the
name of it be?
Original comment by jpanzer@google.com
on 26 Feb 2010 at 5:53
I would vote for collapsing them... "replies" i think works suitably well for
both mentions & replies. (perhaps distinguishable by presence or value of atom
threading data?)
Original comment by wal...@gmail.com
on 26 Feb 2010 at 6:02
"upstream" or "salmon-upstream"
because "Salmon swim upstream!"
It shouldn't matter what the type or usage of the salmon is,
just as long as it is a salmon.
You could define by convention that if you "mention" a
specific URI, you could do lrdd discovery (WebFinger for
email address like identifiers) to see if there is a salmon
relation type defined for that URI. If so you could choose
to send the salmon there in addition to any other salmon
endpoints you discovered or already know about from other
sources.
Original comment by mail.ton...@gmail.com
on 26 Feb 2010 at 6:12
#5: +1
Original comment by hjfre...@google.com
on 26 Feb 2010 at 7:18
Any conclusions on this?
Original comment by hjfre...@google.com
on 25 Mar 2010 at 5:43
Based on this feedback, by EOW, I'm planning to modify the "salmon-mention" and
"salmon-reply" endpoints
to just be "salmon" to handle the generic case of "here's a signed salmon you
may be interested in, based on
metadata or content, please process it and do what you will with it". If later
on there are specific endpoints
other than salmon-signer which crop up, they can get similarly specific names.
Please yell if you see any problems with this.
Original comment by jpanzer@google.com
on 6 Apr 2010 at 10:01
No problems, but I like "salmon-upstream" better :)
Original comment by hjfre...@google.com
on 8 Apr 2010 at 11:09
Collapsed salmon-replies and salmon-mentions into one link relation (which
appears in two
typically disjoint contexts: an Atom feed and a user XRD).
However, in a fine example of conservation of link relations, I realized that
this is also
an opportunity to fix another open issue. I've added a link relation
"mentions" to allow
salmon generators to indicate which specific people the user intended to
@-mention in their
content. This also helps disambiguate situations where we end up with text
like "Hey @Bob"
with no context about what domain Bob is supposed to be on. The link lets you
know that
there was an intentional @-mention and not just accidental text. E.g., if you
see a link
rel=".../mentions" href="acct:bobj@example.org" you can mostly ignore the
textual content
for purposes of notification. (If you want to linkify @Bob it gets a bit
tricker and I'm
looking for a least-effort approach for that.)
Original comment by jpanzer@google.com
on 13 Apr 2010 at 9:25
Just to be clear - with the collapsed relations - will you introduce a
"mention" link that is either SHOULD / MUST
for mentions? Currently (as I have mentioned in the past), we are using an
"ostatus:attention" relation to serve
this purpose. We would happily drop this if it was included as part of Salmon
proper. We use it for the exact
purpose that you've outlined in #10.
Original comment by wal...@gmail.com
on 28 May 2010 at 2:34
Fixed in revision 104.
Original comment by jpanzer@google.com
on 19 Jun 2010 at 7:39
Original comment by jpanzer@google.com
on 19 Jun 2010 at 7:39
Original issue reported on code.google.com by
hjfre...@google.com
on 26 Feb 2010 at 1:05