Monday, February 02, 2009

Using Bloggers undocumented label (category) feeds and Yahoo Pipes to create a tech opinion feed out of Gordon's Notes

[3/14/09: See Update below. I loved how this worked, but I had to give up on it. If blogs survive, maybe Google will add Pipes-like features properly.]

Dang. What's a geek got to do to be fully informed these days?

I've only now learned that my 2nd biggest Blogger feature request, feeds on Blogger "labels" (aka categories, sometimes called tags), has been unofficially available for at least a year ...

Blogspot ... Feeds by Category / search labels

Here's the default URL for any search label in Blogger:

xxx.blogspot.com/search/label/labelname

To convert that search label into a Blogger ... feed, replace the string /search/label/ with /feeds/posts/default/-/ so the new feed URL for that search label has the following syntax:

xxx.blogspot.com/feeds/posts/default/-/labelname

For this to work, you should have enabled Site Feeds for that blogs. Blogger Label Feeds will inherit the full or partial text settings from the main feed itself.

A year! I could have easily solved the problem I flailed at only 5 months ago.

The problem is I my opinion blog, Gordon's Notes mixes both technical and liberal political opinions. The technical opinions might fit in with tech.kateva.org but the political opinions won't. In any case, I like to keep this site to the apolitical "facts", so even technology opinions are a bit out.

Now I can organize feeds however I like, using this undocumented feature and Yahoo Pipes.

To start with, here's the label for "technology"

http://notes.kateva.org/search/label/technology

and here's the feed (Atom) following the above pattern

http://notes.kateva.org/feeds/posts/default/-/technology

It's a well formed feed, except that it doesn't get a useful name like "Gordon's Notes - technology"; it just gets called "Gordon's Notes".

That's good, now we add Yahoo Pipes (best thing they've ever done).

At Yahoo Pipes I can merge multiple GN opinion feeds (Case matters, spaces are sort of ok but, you know, spaces are always a pain)

http://notes.kateva.org/feeds/posts/default/-/Apple
http://notes.kateva.org/feeds/posts/default/-/data lock
http://notes.kateva.org/feeds/posts/default/-/drm
http://notes.kateva.org/feeds/posts/default/-/Google
http://notes.kateva.org/feeds/posts/default/-/Google Apps
http://notes.kateva.org/feeds/posts/default/-/netbook
http://notes.kateva.org/feeds/posts/default/-/os x
http://notes.kateva.org/feeds/posts/default/-/palm
http://notes.kateva.org/feeds/posts/default/-/spam
http://notes.kateva.org/feeds/posts/default/-/technology
http://notes.kateva.org/feeds/posts/default/-/xanadu

The feed summation is pretty trivial ...

image

After I save and rename the YP feed I end up with:

http://pipes.yahoo.com/jfaughnan/gn_tech

Google Reader treats this as a proper feed. So, after years of blogging, I'm now able to do my own syndication.

This will be fun.

Update 2/7/09: Since I had to treat each label as a separate feed, and since a post can have multiple labels, the above design produced a lot of duplicates. It also had no sort order. I added the "Unique" (by ID) and "Sort" (pub date) operations to fix those problems.

I've also created a new feed that takes the the pipe for http://pipes.yahoo.com/jfaughnan/gn_tech and then adds (Union) the feed from Gordon's Tech (this blog) and sorts by publication date. I didn't need another unique operator because I wasn't creating more duplicates.

This new feed combines all my posts from Gordon's Tech with my technology opinions posted in Gordon's Notes. It's

http://pipes.yahoo.com/jfaughnan/tech_plus
and the new feed is
http://pipes.yahoo.com/pipes/pipe.run?_id=CGBvCFr13RG4ekbA1JzWFw&_render=rss
I might add this to my post footer but I've noticed two limitations of this type of feed.
  • The posts sort correctly when I view the generated Pipe web view, but the feed from that view gives all posts close to the same pub date.
  • The feed does not inherit the title set in properties, it gets the URL string.
Update 3/8/09: hmm. There might be a serious problem here. Lately Google isn't indexing my stuff as well as they used to. I've seen that before when Google gets confused by splogs. It indexes splogs that have harvested my posts, then it starts treating my blog as though it were also a splog. Usually it self-corrects after a few days but this time it seems to persist.

Alas, by creating Pipes from my feeds I wonder if I've created my own de facto splogs. I tried to submit a question to Yahoo about making these non-indexed but the submit forum crashed. I think I'll have to take these offline. Unfortunately it's very hard to learn Google's policy on these things.

Update 3/15/09: I never did get an answer on my Yahoo or Google Groups questions. I did find that I was getting odd results from subscribing to the Pipes I created. Google Reader persistently showed me old posts as new, probably because when a Yahoo Pipe is generated all the source posts get the same date.

In addition, even though I "unpublished" the Pipes I linked to them from various posts, and I think Google was finding and indexing them. I don't see how Google could differentiate them from spam blogs, and we know that the GoogleMind associates splog taint with the parasitized victim. (The GoogleMind is a harsh and unforgiving Yahwhite god.)

Yahoo Pipes needs a no-index tag and it needs to handle post dates differently.

I didn't want to delete my Pipes in case I wanted to work with them in the future. So I simply changed the URLs to something new and thus invisible.

No comments:

Post a Comment