I assume you mean other than the way it seems to have shut down on August 14th?
I actually wrote a seven-paragraph discussion of pitfalls and things that would be needed to make it work, and then I came back and remembered Bob Wyman saying earlier today that PubSub.com sees an average of 24 updates per second. I’ve got a changes.xml file here with 100 items, it’s 11.86KB. If I’m counting on my fingers right, that means that changes.xml for 60 minutes (the minimum needed to be useful to an aggregator) would amount to 11.67GB. Wanna start your aggregator run by downloading 11.67GB, before you even start to pull updated feeds? Compress it like crazy, of course, you’ll still be able to tell when your aggregator starts to update, as first you connection gets saturated, then your CPU spikes.
If it was a matter of someone with bandwidth and CPU to spare, who already knows all and sees all, allowing you to register a list of URLs of interest, and then either telling you which ones on your list it knows changed, or while it’s at it, giving you the entries, well, you’d have Bloglines 🙂
Matt: What timezone is that? Does it include daylight savings time?
Interesting idea though. Personally I prefer the blo.gs approach, where you tell a central service which sites you are interested in and it provides a feed of update times. Central points of failure are bad though, so sort out a decentralised system where anyone can run a notification server and the quality of information is protected by digital signing of some sort. Then watch the whole thing collapse under people’s inability to use encryption software…
I assume you mean other than the way it seems to have shut down on August 14th?
I actually wrote a seven-paragraph discussion of pitfalls and things that would be needed to make it work, and then I came back and remembered Bob Wyman saying earlier today that PubSub.com sees an average of 24 updates per second. I’ve got a changes.xml file here with 100 items, it’s 11.86KB. If I’m counting on my fingers right, that means that changes.xml for 60 minutes (the minimum needed to be useful to an aggregator) would amount to 11.67GB. Wanna start your aggregator run by downloading 11.67GB, before you even start to pull updated feeds? Compress it like crazy, of course, you’ll still be able to tell when your aggregator starts to update, as first you connection gets saturated, then your CPU spikes.
If it was a matter of someone with bandwidth and CPU to spare, who already knows all and sees all, allowing you to register a list of URLs of interest, and then either telling you which ones on your list it knows changed, or while it’s at it, giving you the entries, well, you’d have Bloglines 🙂
Phil,
http://pingomatic.com/updated/http://photomatt.net/
Whatcha think?
Matt: What timezone is that? Does it include daylight savings time?
Interesting idea though. Personally I prefer the blo.gs approach, where you tell a central service which sites you are interested in and it provides a feed of update times. Central points of failure are bad though, so sort out a decentralised system where anyone can run a notification server and the quality of information is protected by digital signing of some sort. Then watch the whole thing collapse under people’s inability to use encryption software…
Simon, GMT of course. I’m not tied to the format, it can be anything.