Mark Nottingham wrote:
Your arse-kicking aggregator should appear in the new ~aggregator's~ RSS feed just one time .... hopefully all the B's will pick it up at that time. Should they not, you should be able to go to any B and register it again.The registry does not need to be centralized. foo can be wherever, it can vary from blog to blog. A centralized registry is a natural thing here, but is by no means a monopoly .... anybody could setup the script to run on their own server and the list of aggregators and their corresponding URI formats could be circulated via RSS.So, site A has a feed that I'm interested in, and they use service B to dispatch subscriptions. B knows about aggregators 1,2 and 3, but how does it know about my own, hand-written, arse-kicking aggregator, Q?
Nope they don't need to match up the feeds to the aggregators because the input post contains that matching: <a href="fooB?action=subscribe&feed=BlogA.rss">common button</a> It just dawned on me that fooB ~should~ be able to pick up the feed via the referal, but in practice it probably can't so that it would be included in the post variables.Of course, you could have a register-your-aggregator service as well, so that B becomes a sort of marketplace of feeds and aggregators, matchingthem up as appropriate.
The fooB kind of script pages don't need to maintain a database of feeds because that is input to them. They do have to maintain a list of aggregators and the URI format for each. The user's cookie for fooB contains the other variable which is what their chosen aggregators are. Note I said aggregator(s), no reason why you couldn not have more than one.However, as a user, I still have to go to B, register myself as a user there, add my aggregator if they don't have it, and hope that they have all of my feeds; otherwise, I have to manuallyenter them.
Nightmare. What's the point?This is not really as bad as you are making it out. There are far more feeds than there are aggregators. The aggregators should be able to create a cloud of so that any new aggregator would be picked up in a matter of hours. For the end user it is a one time selection and from then on they just click on the one recognizable button to subscribe to whatever they find on the wild web.
I don't understand this problem. I have an aggregator scritp running on my localhost:2187 and it is behind a firewall. I have no trouble in clicking on buttons that i put up on the web and having those buttons post to my local process. Could you give me more specifics ?Additionally, it still doesn't address the case where my aggregator is behind a firewall, and can't expose anything for B to talk to.
Yep, only problem is that it will be years before all the browsers support this kind of thing ... or am I missing something.Using the Content-Type to dispatch a particular format (RSS in this case) is so much simpler, robust, and - more to the point - it's how the Web is designed to work.
Seth Russell