Jeff
On Sun, Feb 27, 2011 at 12:44 PM, Jeff Lindsay <[email protected]> wrote:
>> The objection to this boils down to the difficulty of setting up the
>> publisher-hub relationship, compared to everything that you get for
>> free with feeds.
>
> I'd like to start thinking about this again ... any pointers to
> previous discussions on this?
IIRC these discussions were in Aug/Sep 2009.
> It's also worth noting that dropping federation and considering in
> practice most people implement their own hubs as providers, this
> drastically simplifies the model. You lose the need for publisher-hub
> relationship, you better match people's models of how it should work,
> etc.
By federation I really mean chaining, whereby the same protocol ("-->") enables:
P --> H --> H --> ... --> H --> S
So, yes. You could also have P --> S.
alexis
> I think previously the assumption that publisher would not want to run
> a hub because they have feeds might need to be revisited with new data
> suggesting people interested in webhooks/event APIs are willing to run
> software that would handle the responsibilities of the hub.
>
> Perhaps what comes out of this is something much less grand as the
> original vision for PSHB (federated hubs, etc), but something much
> more practical and simple (as another spec or what, I don't know).
>
> -jeff
>
>>
>>
>>
>>
>>
>> On Sun, Feb 27, 2011 at 9:28 AM, Daniel Chapman <[email protected]>
>> wrote:
>>>> Either way, there are common issues that overlap.
>>>
>>> Absolutely. Which is why I suggested that PSHB define "new push project X"
>>> as the mechanism by which it transports feeds.
>>> The main downside to that though, is that there wouldn't be a lot left to
>>> talk about with Atom and the Push in separate projects. PSHB would just
>>> become "Push project X" + "Atom Feeds".
>>> My main concern with generalizing PSHB is that then you can't rely on a
>>> rel=hub link to be a real feed hub. PSHB + Atom is a bit confusing to put on
>>> a link on your site don't you think?
>>> But if we started with a new spec for the push, then the expectation right
>>> from the start is that new PPX endpoints and pushes are arbitrary and not
>>> for blind consumption.
>>> It just feels like that expanding the spec to include anything, takes the
>>> currently nice trustable rel=hub links and makes them suddenly uncertain. I
>>> am not sure that is a good thing.
>>> A PSHB -Atom +JSON (or +whatever) appendix is possible, but just seems a bit
>>> messy.
>>
>
>
>
> --
> Jeff Lindsay
> http://progrium.com
>
