I don't think this is within the scope of Atom or any other spec. If you want to differentiate your others, why not invest some time in a decent duplicate detection algorithm like Google did? This is your problem.


Graham

On 7 Apr 2005, at 7:29 pm, Bob Wyman wrote:


Robert Sayre wrote:
Establishing equivalence only addresses a part of the problem.
Fully agree. I just wanted to point out that a part of the problem is
more solved than your post indicated.
My apologies if I made the situation sound more dire than it is.
However, even with the "alternate" we've still got a pretty big problem.


HTTP/1.x 301 Moved Permanently
Another partial solution... Useful in some situations, but relying
on it prevents anticipating change and thus may result in some data loss.
Also, a change like this requires access to the web server configuration in
many installations and hosted users can't always get such access.


[When there are multiple feeds] Some aggregators continue to poll both
locations, though. Not sure why.
If they're doing it on purpose, I'm not sure they'd pay attention to
feed links either.
The automated aggregators are stuck polling all feeds in many cases
because:
1. They can't determine the equivelance between feeds without
explicit statements from the publisher.
2. Even if equivalence is known, we can't know which feed is
"authoritative" or if one feed will be deleted in favor of the other. Thus,
we've got to read both to ensure that we are reading which ever one is more
long-lived.
3. We have to read all feeds since users like to be able to
subscribe to feeds using file names or masks on file names... If we don't
read all feeds, we will incorrectly fail to deliver to the user content that
we have available to us.


I'm not sure they'd pay attention to feed links either
        I *guarantee* you that if we had an accepted way to address this in
feeds, PubSub *would* pay careful attention to it.

                bob wyman




Reply via email to