My only real issue was about feeds having to be active and pinging hubs. After reading a discussion that hubs could be designed to do a manual fetch of the feed and that it wouldn’t be against protocol. You couldn’t take advantage of the real-time aspects but your not shut out.

So I spent some time looking over the spec but I never learn anything without playing around. So on the last day of my vacation I managed to eek out some time and start playing. First thing was to build an end point and have it email me the request as I manually subscribed to the hub at http://pubsubhubbub.appspot.com. Next day I got the end point to respond with the right challenge phrase, and now subscriber pings were flowing.

I sat on that for a bit trying to get a feel for the pings and subscribed to a few more feeds. I think it was at this time I realized the greatest thing about pubsubhubbub, I didn’t have to build a feed engine, just a catchers mitt. Sure for what I had in mind I’d have to slice up the <item> pretty deep, but I don’t have to manage a threading model to efficiently consume feeds. Yippeeee! Awesome.

So now I have an end point that does about 75% of the <item> parsing that I want to do and 7 feeds with two of them being google shared feeds.

You can take a look at a quick little reader I built at http://http://punchingsoup.com/#pubsubhubbub_panel. It’s ugly and doesn’t work quite right but it’s good enough to show.

Userscripts.org

Unfortunately I didn’t have time to build in a way to store search topics in cookies, maybe next version.

I was surprised at the lack of search operators. I expected to be able to search just the title or return only unread posts.

Here’s how to modify the script to add your own search topics.
Open the script up and all you’ll need to modify is this section.

var smarttopics = new Array;
smarttopics = {
"Buzz" : "leak|deadpool|acquisition|rumor|layoff|merger",
"Apple" : "apple|iphone",
"Microsoft" : "microsoft|windows"
};

Say you want to add a topic called Sun and it searches for Sun and Java. Add the following line above the “Apple”… line.

“Sun” : “sun|java”,

Removing a topic is just deleting the line with that topic on it. The biggest thing you’ll need to watch is the comma’s, make sure each line except the last has a comma after it.

It’s still pretty much a toy project and my first in python, but it is being used on http://punchingsoup.com and run every hour.

rss2db.zip

The setup is pretty simple.
1. Run rss2db.sql on your mysql server.
2. Modify the connection information to that mysql server in rss2db.py.
3. Run rss2db.py.

From here i’ll start building out a php based web frontend for the database that mimics what i’ve built in .net for myself.

Should be fun.

I’ve annotated the screenshots on Flickr. You can view those here RSS Reader Set

Get it here, friendfeed_on_twitter.user.js

Here’s just the search bar script, twitter_search.user.js

I also found out that Friendfeed updated their api with a way to query for a url. So i built a greasemonkey script to add a link to friendfeed on every url. I couldn’t figure out how to put the icon after the link and javascript isn’t something I’m an expert at. So far the url search only works via the api and not on the search part of Friendfeed, so the icon opens a window with the xml output of the api call. I fix this when you can search for a url on the main Friendfeed interface.

Get it here, friendfeed_link_on_twitt.user.js

Be kind these are the first greasemonkey scripts I’ve made.

So after about 5 minutes of surfing and passing by shareware products, I came across this url, Decrypt encrypted stored procedures.

I took the stored procedure and turned it into a script that would automatically decrypt the stored procedures in a database. One thing this can’t handle is decrypting stored procedures that are > 4000 characters, since they take up more the 1 row in syscomments. So for me I was able to decrypt about half of the 1300 stored procedures this application uses. Also about half the decrypted stored procedures were recreated under the id I was running the script with rather then dbo.

The logic is pretty simple to follow so I think I might spend some time and build a .net app to decrypt the stored procedures that are > 4000 characters. Don’t hold your breath though, with the stored procedures I decrypted and profiler I was figure out what I needed to figure out.

Download: decrypt.vbs

I’ve added an integer field to my feeds table so I can individually define the time to live for the articles in each feed or 0 for forever. I also added a field to each article to flag if it was faded or not for reporting.

Now a script runs nightly to fade articles and keep my unread list trimmed down. This keeps me from having to check feeds with a large number of unread articles and manually clearing them out.

As it stands now a bundle is a group of tags and feeds that have some common focus. I’ve got one for mobile that includes feeds from wapreview, MobileCrunch, SMS Text News and a few others, plus the tags: mobile, phone, verizon, sprint, tmobile, sms, smartphone and a few others. This way I can follow the mobile industry not just via mobile focused blogs, but also articles that are tagged with a mobile focus.

As long as the bundles stay under 50 articles a river style reading will do fine. If the bundles consistantly have more the 50 articles then I may have to include a secondary grouping structure probably in the form of a tag cloud.

Tag Cloud – This opens a tag cloud that shows the current tags associated with unread articles. After I’m finished reading the articles directly associated with a chosen tag, I’m prompted to search the rest of the unread articles for that tag in their title. This is useful for tags like Microsoft or Google but not so much for tags like product profiles used by techcrunch or daily trivia.

Feed List – This shows a list of all the feeds that have a weight greater then 1 or have 5+ articles. With 100 feeds showing them all is to long a list so it’s geared towards the ones I want to hear from the most. I also have special feeds for articles that have diggs and for articles flagged as Techmeme.

Latest – This ones pretty simple and shows articles with the latest at the top.

Oldest – Same as latest but with the oldest at the top.

Review – List of the articles flagged for review.

Recent – Most recent articles read in reverse chronological order. Helps when I accidently mark something read.

Search – Can search the title of the article with the option of including the articles body also.

Admin – Used to manage feeds and some basic reporting.