is really "steve jobs" part of apple stock value...? designers, engineers, products, marketing strategies are.
- ezekiel
He is a part of the stock value. All CEOs should be. If they aren't, find a new CEO. Until Apple comes out with their strategy to succeed Jobs and prove they can still innovate, he's still a very valuable part of the company. Apple needs to use this leave to prove Apple != Jobs.
- ·[▪_▪]·
"If your application uses the FriendFeed API, you can now show FriendFeed updates to your users in real-time using the new FriendFeed Real-time API. We're providing access to the same underlying technology that's used by the real-time view on FriendFeed to all FriendFeed developers."
- Bret Taylor
from Bookmarklet
This is the gamechanger for FriendFeed - just wait and see
- Ed Dale
sure it is! but hope friendfeed can scale better than twitter...
- Aziz
Bret, IMHO, instead of providing such "fantasy" features for the API, you should raise the limits you told me that you weren't able to raise some months ago... But, heh, you decide :)
- directeur
from NoiseRiver
Nice work guys. Keep up the innovation and stay focused.
- Louis Gray
directeur: mind sending an mail to api@friendfeed.com detailing what limits you are referring to? Sorry if we dropped something - it was likely just forgotten due to something else, but we can address your problem now with all the details.
- Bret Taylor
Nice job. I would have preferred that the term "Real-Time" be reserved for an XMPP like implementation. As of now this feature is great but it is not exactly real-time. It is polling after all. This is a massive improvement though. Great work FF crew. Keep pushing the envelope. Don't fear the fire hose!
- Rolf Schewe
Bret: Thanks for the answer, I'll send you an email ASAP with the details! Again thanks for the attention
- directeur
from NoiseRiver
Rolf, this API supports "long polling" -- it's every bit as realtime as XMPP.
- Paul Buchheit
@Rolf: with long polling, you get updates as soon as we have them, so it is as real-time as XMPP, just a different delivery mechanism.
- Bret Taylor
Bret, what's real-time in the case of Friendfeed? Really? I still have the picture of an axajy widget that refreshes and sends queries... It's pulling from the server, not the other way, right?
- directeur
from NoiseRiver
Gentlemen, Friendfeed 1.0 is a cool service. With RealTime API, I think we're looking at "killer app" with a huge breadth of applications. Well done.
- John Craft
So it is also dependent on the delivery on the other end I suppose. I just don't get why I get some content fast such as Twitter and some after a long delay such as Google Reader shared items.
- Rolf Schewe
Google Reader shared items are definitely not real-time.
- Rolf Schewe
From a community building perspective - this real-time is more than enough. If your a programmer designing an app with the API - here is the NUMBER 1 request - Make sure your app doesn't jump out of the stream when you comment or like something
- Ed Dale
I added Facebook today and tested to see the delay. The status message came through 1 hour later. I actually timed it.
- Rolf Schewe
@rolf: ff servers have to decide how often to crawl the feeds; they don't have the luxury of a realtime api to all of those services. if you post something to ff directly, however, you'll notice it show up immediately
- Karl Rosaen
Rolf- As Karl says, for incoming content we're at the mercy of the external providers. Our proposal for solving that side of the problem is called SUP: http://blog.friendfeed.com/2008...
- Casey Muller
Casey, suppose I write an entry and delete it just after. Will it still be in the ral-time view for other users?
- directeur
from NoiseRiver
it will increasingly be helpful to know which feeds FF does have realtime access to
- Steve Gillmor
So in essence the real-time feature is in some terms conditional on the outside services. Kind-of real-time.
- Rolf Schewe
help us to put pressure on them by listing those who have XMPP pipes in
- Steve Gillmor
directeur- Yes. Anybody (including API clients) monitoring the real-time feed will already have received the entry before you deleted it
- Casey Muller
casey can you guys publish a list of inputs with polling frequency please
- Steve Gillmor
The real innovation (sorry to insist on this) will be to process feeds on the user's level. Attention is the keyword. I DEFINITELY don't care about real-time. I want intersting stuff not fastly delivered ones... just sayin' :)
- directeur
from NoiseRiver
Steve, Twitter is the only service that sends updates by XMPP. All other services (except things posted directly on FriendFeed or via the API) arrive via RSS/Atom feeds, which are generally checked once every 1/2 hour. However, we also monitor several pinging services and utilize SUP to discover feed updates much faster and can often get new entries within a minute or two.
- Paul Buchheit
out of curiosity, what does the ff realtime view set the timeout to? (i should clarify: I'm talking about calls from the client JS to the realtime APIs; what's the timeout of those requests)
- Karl Rosaen
Karl, it's currently about 60 sec, but that's mainly due to concern about browser and proxy timeouts. The server responds as soon as there is an update, so there's no disadvantage to using a larger timeout.
- Paul Buchheit
@Paul, thanks, i was wondering for a comparison to the timeout you'd need to use if supporting such an api from an app engine app (where you have to respond within 10 seconds or so I think)
- Karl Rosaen
Thanks for the feedback and the links. It helped a lot. Maybe down the line we will be able to get RSS data in a little faster. The Google Reader floods can get a little heavy with the current polling interval.
- Rolf Schewe
Paul would other sources who provide XMPP be accepted? Or is Twitter exclusive?
- Steve Gillmor
Steve: We would be happy to support more XMPP services depending on the ease of implementation. It is certainly not exclusive to Twitter.
- Bret Taylor
Interestig to see how adding DOM operations evens out John's benchmarks (first has DOM ops and Safari beats Chrome, second is Google's apparently slightly biased benchmark, which relies heavily on recursion).
- Bret Taylor
from Bookmarklet
It is time to optimize for DOM. Faster JavaScript is fantastic..... great for games and crypto.... but what about the real world? NOTE: Ray Cromwell's Chronoscope benchmark http://timepedia.blogspot.com/2008...
- Dion Almaer
These tests are very insightful. I was initially very excited about Chrome being 10 times faster on javascript. As usual with any such claims (not just from Google), they are skewed. However, I do think that Chrome entering the market will cause all browsers to get better more rapidly. Competition does drive innovation.
- Robert Felty
Thanks for sharing this detailed knowledge. Very useful.
- Edwin Khodabakchian
we're working on DOM work now. believe that the tracing technique has applications; working on getting the DOM acting faster soon.
- John Lilly
granted all beta-quality of Chrome and zero-day securiity bug, i think their current *speed* shows only lack of exception handling. Once it matures, it will be same or evenn worse than FF - it is tough to beat long-standing code without rounding corners on usual code glue...
- A.T.