generate feeds
Ticket
+solution
What do we want feeds for? Initially:
- Search cards (pretty much gets Recent Changes)
Eventually:
- The watchlist (see implement notification system)
- Individual pages' Changes
- Pointers?
- ??
Helpful references/tools:
Atom - http://www.atomenabled.org/developers/
RSS - http://www.rssboard.org/rss-specification
convert RSS to Atom - http://atom.geekhood.net/
http://feedvalidator.org/ (works for RSS and Atom)
Feed autodiscovery - best practice is probably to only put in one (Atom or RSS). See http://philringnalda.com/rfc/draft-ietf-atompub-autodiscovery-01.html
Ruby tools for RSS & Atom:
This is really both "testing" and "in progress", since there are a couple of loose ends to clean up, but we're wanting to start getting feedback on what we've done so far.
http://wagn.org/wagn/*recent_changes.rss looks okay with a quick look in Google Reader.
One missing thing - there's a way to specify the URL of the web page associated with the feed (i.e. http://wagn.org/recent ) - and it looks like this is broken in our feed (in the Google Reader interface it shows up as a double >> symbol just to the right of the title of the feed)
After looking at the validator, i think we have /wagn/*recent_changes and it needs to be a full URL. Also, suggest http://wagndomainname/recent rather than above. --John, Feb 10
--John Abbe.....Sat Feb 07 00:28:05 -0800 2009
The validator has some suggestions: http://feedvalidator.org/check.cgi?url=http%3A%2F%2Fwagn.org%2Fwagn%2F*recent_changes.rss
Full URLs and spelling description correctly seem like winners :-)
The other suggestions look trivial and like a good idea too.
--John Abbe.....Tue Feb 10 16:32:57 -0800 2009
I fixed description, made primary urls full. I tried adding the atom:link tag suggested by the validator but it broke the feed in my Safari, so I left it out. Urls in the body/description are a little tricker to fix-- should be done but needs a little more digging in.
Also, while looking at this I found myself wondering if we should do atom instead of rss. atom is newer format with a number of advantages. rss2.0 is the conservative choice. google, firefox etc. seem to be pushing atom, it seems likely to me that's where we'll want to be in future.
--Lewis Hoffman.....Thu Feb 12 18:19:13 -0800 2009
do we have to do one or the other?
--Ethan McCutchen.....Fri Feb 13 08:48:52 -0800 2009
We can do both, but in terms of auto-discovery & communication I think it will be less confusing to just have one. Otherwise everytime you talk about a feed from a wagn you have to specify which one you're talking about. I think I messed up by not considering this more carefully early on, and I'm now thinking we should go with atom. Happily, I think it's an easy change (hard work here was dealing w/ transclusions). Also need to handle full urls throughout the body. Maybe we want to get some other opinions on it..
--Lewis Hoffman.....Fri Feb 13 10:18:57 -0800 2009
Cool!
I'll check in a couple more feed readers, but my guess is the relative URLs can wait for post-1.0.
Atom was created to be more precise and complete than RSS 2.0, and has some additional features built in, so i'm in favor of that, and unless we decide otherwise, will make a ticket for that when i close this one.
--John Abbe.....Fri Feb 13 23:14:49 -0800 2009
--John Abbe.....Tue Feb 17 22:48:04 -0800 2009
Documentation done except for something short about feed readers.
--John Abbe.....Thu Feb 19 19:05:02 -0800 2009
Whup, just found that it's not working on Searches - clicking the feed icon from http://wagn.org/search/demos gets http://wagn.org/wagn/*search.rss which gets a 500 error.
I played around a bit looking for what the URL is supposed to be but i don't think there is one. http://wagn.org/search/demos.rss was amusing though :-)
Suggest closing this and ticketing generate feed and feed URL for searches.
--John Abbe.....Fri Feb 27 16:28:32 -0800 2009
I think we ought to peek and see if it's a quick fix.
--Ethan McCutchen.....Fri Feb 27 17:02:16 -0800 2009
Documentation done, and ideas considered but not implemented added to design list.
--John Abbe.....Sun Mar 01 00:51:42 -0800 2009
I think I got searches working
--Lewis Hoffman.....Mon Mar 09 14:06:12 -0700 2009
Using http://wagn.org/wagn/Blog_entries_by_date (signed out) as a test case. Confused about pubDate. When i look at the feed in Firefox (by just clicking the feed icon), i get the items in the order they are in the .xml file. When i look at the feed through NetNewsWire, they're in pubDate order, which is different. This is not a bug, just a preference that each feed reader may have.
The take-away for us i think is to try to not update old blog entries unnecessarily.
--John Abbe.....Mon Mar 09 23:18:49 -0700 2009
Most of the links in this discussion are broken. JF is interested in rss for TT wagn, so I need to know better how it works and what is there.
--Gerry Gleason.....Fri Nov 06 12:21:59 -0800 2009
Hmmm, the automatic dates (created and edited timestamps that AR creates and updates for us) may not be sufficient to implement blog semantics. Probably should give entries an explicite 'pub-date' in a plus card to the blog entry cards, and use that to format permalinks. We need to separate the production workflow (how a blog entry or other published item is created by an Editor) from the publication process (Adding it to the public blog and feeds), and editing (fixing typos, renaming entries, etc.). Failing that, the creation datetime needs to be the source for permalinks.
Copying your comment here to blogging, Gerry.