Thoughts on FeedFeeder
Nov 5th, 2006 by Jon Stahl
I’ve been checking out Zest Software’s new product FeedFeeder. It’s pretty darn cool. Basically, it’s a simple ATOM aggregator for Plone… with a twist: it turns the aggregated items into first-class Plone content objects.
This is big. What it means is that you can finally use Plone as a real content aggregator, and store the content for long-term archiving, remixing, etc.
Of course there are some obvious rough edges. (That’s ok, it’s still in alpha.) Here’s what I saw.
- FeedFeeder only parses ATOM feeds. To me, this seems like a pretty major barrier to a lot of real-world use, since ATOM isn’t nearly as widespread as it deserves to be, ahd there’s still a lot of RSS 2.0 out there, and since feedparser handles
it, I don’t see why FeedFeeder shouldn’t be able to cope pretty easily. - Right now, you’d need to manually set up a cron job in order to auto-refresh feeds. I know this is something of a Zope limitation, but it’s still kind of a barrier to deployment. I wonder if it is possible to implement something along the lines of PloneRSS and set a “freshness time”, and update when the page is called. Crude, but perhaps user-friendly. Or else we could take advantage of Zope 2.10’s clockserver (but not until Plone 3.0).
- It would be really handy to be able to choose the default workflow state for aggregated items. Right now, it appears to set everything to public draft. But a site administrator might want to have everything published, everything private or to put stuff into a custom workflow state for reviewing.
- Couple of minor nits:
- The list of aggregated items appears to show dates in European format (dd-mm), rather than obeying the date format defined in site_properties.
- The list of aggregated items links directly back to the original, rather than to the copy in Plone. That seems counter-intuitive.
- FeedItems could use an icon that is not a folder.
I can’t wait to see FeedFeeder continue to mature. Big cheers to Jean-Paul, Rocky and their fellow Zest-ians!
UPDATE: As usual, I am a day late and a dollar short. A lot of this stuff was worked on at the recent Seattle Sprint. Not merged back to the trunk yet, but that should happen presently. And they have some good roadmap ideas, too!
Thanks
RSS support should be easy. Feedparser (which does the heavy lifting) parses it alright, we just might need to grab a different fieldname or compensate for a missing ID or so.
There are one more issue: I’m a bit worried about large sets of items myself. Because of the extensibility we effectively need to wake up every object to see if it implements some interface. That needs a catalog field for it to scale. I know there’s some generic catalog solution being discussed.
But all in all, it seems to work real well. At least for our original client (which is the reason behind the strange date format, that should change).
Thanks a lot for this feedback about feedfeeder.
Some of the issues you mentioned have actually been fixed during the after conference sprint, but still resides in a branch:
Select target workflow state for feed items.
Links point to the Plone item instead of the original item on the Feed listing and a link to the source is added to the feed item view.
Date format changed to Plone’s default way, that was a left over of the initial implementation for our customer.
Just to let you know. These changes have now been merged to trunk and will be available in the official release, with many thanks to the people that helped during the seattle sprint.