Question on creating RSS feeds

[Stefan Kurla <st...@gmail.com>]

I would like to create RSS feeds and email notifications for my
project. I have seen it in action on infoQ (a site that I particularly
like, before I learnt that it was built on jackrabbit).

There is the observation mechanism in Jackrabbit that consists of
creating event listeners on node paths or the workspace in general and
these event listeners are created on node add and other events.

In my case, I expect people to want to listen to events raised on
documents being added to the repository in general. Also, I expect
users to want to be aware of changes being made to their subtrees.
They could be notified by RSS or email. Now I have not worked on
either observation or creating an RSS feed so this is kind of chicken
and egg for me. Might be really obvious to someone else.

I believe that if I have 100 users and each of them listen to 10
different node paths each, I have a lot of listening for events going
on. Would that adversly effect the repository in terms of performance
and constant event processing. I guess is the approach of listening
for events and processing events as they are raised the right way to
go?

Am looking for inputs and suggestions.

Regards.

Re: Question on creating RSS feeds

[Stefan Kurla <st...@gmail.com>]

>
> > When building the feed support for InfoQ, I was thinking in terms of
> > pull vs push feeds. I am not sure if these terms are usually used in
> > conjunction with feeds, but what I mean is: is it possible to
> > pregenerate my feed so that once I get a request this is already
> > prepared (push model) or should I generate the feed each time I am
> > receiving a request (pull model). Having these questions answered,
> > I've been able to further implement things.

AP=> What model did you choose?


>
> I'm planning also this feature for our application. Actually I have not
> decided how to implement it, but I think the RSS will be generated on
> each request because not all users have access to all the documents (Out
> application is an document manager).

That is the problem that i am facing as well. my feeds are going to be
very personalized, one could be looking for events at
/jcr:root/myFolder for changes the other could look at
/jcr:root/myFolder2 for changes. If this were the case then I would
have 2 event listeners then.. for 1000 such cases, there would 1000
event listeners. This would be the push model, where as the data is
published the RSS xml is created or an email is sent out as a
notification.

That seems to be the model that i am looking for since we have to
inform the user of the changes in the repository.

Any suggestions?

Re: Question on creating RSS feeds

[Paco Avila <pa...@git.es>]

El jue, 17-05-2007 a las 01:29 +0300, Alexandru Popescu ☀ escribió:

> When building the feed support for InfoQ, I was thinking in terms of
> pull vs push feeds. I am not sure if these terms are usually used in
> conjunction with feeds, but what I mean is: is it possible to
> pregenerate my feed so that once I get a request this is already
> prepared (push model) or should I generate the feed each time I am
> receiving a request (pull model). Having these questions answered,
> I've been able to further implement things.

I'm planning also this feature for our application. Actually I have not
decided how to implement it, but I think the RSS will be generated on
each request because not all users have access to all the documents (Out
application is an document manager).
-- 
Paco Avila <pa...@git.es>

Re: Question on creating RSS feeds

[Alexandru Popescu ☀ <th...@gmail.com>]

On 5/16/07, Stefan Kurla <st...@gmail.com> wrote:
> I would like to create RSS feeds and email notifications for my
> project. I have seen it in action on infoQ (a site that I particularly
> like, before I learnt that it was built on jackrabbit).
>

Thanks for the nice words.

> There is the observation mechanism in Jackrabbit that consists of
> creating event listeners on node paths or the workspace in general and
> these event listeners are created on node add and other events.
>
> In my case, I expect people to want to listen to events raised on
> documents being added to the repository in general. Also, I expect
> users to want to be aware of changes being made to their subtrees.
> They could be notified by RSS or email. Now I have not worked on
> either observation or creating an RSS feed so this is kind of chicken
> and egg for me. Might be really obvious to someone else.
>
> I believe that if I have 100 users and each of them listen to 10
> different node paths each, I have a lot of listening for events going
> on. Would that adversly effect the repository in terms of performance
> and constant event processing. I guess is the approach of listening
> for events and processing events as they are raised the right way to
> go?
>
> Am looking for inputs and suggestions.
>

When building the feed support for InfoQ, I was thinking in terms of
pull vs push feeds. I am not sure if these terms are usually used in
conjunction with feeds, but what I mean is: is it possible to
pregenerate my feed so that once I get a request this is already
prepared (push model) or should I generate the feed each time I am
receiving a request (pull model). Having these questions answered,
I've been able to further implement things.

./alex
--
.w( the_mindstorm )p.



> Regards.
>