Tracking Android Source Branches Via RSS

My recent work on Android has taken me beyond the confines of the usual SDK madness and into the world of the Android source and build system. Unfortunately, things like announcement mailing lists seem to be going the way of the dodo, and I found no straightforward method of tracking branches/point releases. Androidtracker is a small project that tracks Android source branches on an hourly basis and writes them to RSS. This is helpful because minor release announcements are often glossed over, and Android Building is a very busy Google Group to keep up with. The feed is below:

Android Source Tracker

Note: Currently only the initial import of branches is shown, but it should grow over time.

As always, feedback and suggestions are welcome. I can be reached by email, or on Twitter or GitHub.

Hacker News Feeds With Rolling Time Windows, take two

While sound in theory, my previous excursion into frequently-updated RSS feeds fell down in the case of Google Reader caching items, forever. This is for good reason; per the Google Reader Help, they cannot determine whether an item was removed intentionally, or simply fell off the bottom of the feed. This resulted in a Top 10 feed, for instance, having 15 or 20 items by the end of the day if it was refreshed too frequently.

The simple solution to this is to start tracking which articles have been published, and blank out items after they no longer deserve to be Top X material. It doesn't hurt to space out the publish cycle by a few extra minutes, either. Revised feeds are below:

HN Top 10 Stories
HN Top 15 Stories
HN Top 20 Stories
HN Top 25 Stories

HN Top 10 Discussions
HN Top 15 Discussions
HN Top 20 Discussions

HN Top 5 Underrateds
HN Top 10 Underrateds
HN Top 15 Underrateds

As always, feedback and suggestions are welcome. I can be reached by email, or on Twitter or GitHub.

Really Simple RSS Downloader (RSRD)

My latest scratched itch: Really Simple RSS Downloader (RSRD) is a simple downloader for RSS enclosures (attachments). It does not require any high level languages, XML libraries, or databases, and is easily run through cron or at.

Features:
- Downloads RSS enclosures, manually or as scheduled by the user
- Tracks previously downloaded files, but still allows users to delete old files without fear of re-downloading
- Requires only a Bourne-compatible shell and core UNIX tools

Feedback and suggestions are welcome. I can be reached by email, or on Twitter or GitHub.

Comics.com RSS Feeds (Suck)

Most sites want to draw you to their content and provide a plethera of ways to do so. Comics.com appears to not be such a site. Most avid readers of webcomics have moved on from the days of having 15-20 comic bookmarks to using RSS, which polls and present their content for them. For reasons unknown, Comics.com has resisted offering useful site feeds. Feeds do exist, but they have been hobbled in a slew of ways:

A.) Feeds require a login.
B.) All selected comics are combined into a single feed.
C.) Feed item choices don't offer all comics available on the site.
D and E.) Feeds only include the current day of comics and feeds are unique to a user. In a shared refresh system (a la Google Reader), comics can be missed if the individual person's feed is not force-refreshed daily. This could be argued to be a shortcoming of Google Reader, but this is the world we live in.

Comicsdotcomrss is a simple package for building RSS feeds of Comics.com comics content. Configuration is done via a list of tuples at the top of the source.

Update 3/27/11: Comics are now embedded in the feeds, when available.

Update 6/8/11: Comics.com is no more as of last week, and all their content has been merged into GoComics.com. The latter features RSS feeds for all comics, with no registration required. I've taken down the sample RSS feeds below, as there's no longer much use for them, but I will keep the GitHub repo around for posterity.

A few samples (now defunct):
F Minus
Off the Mark
Pearls Before Swine

Note: No Comics.com content is downloaded or redistributed by this framework.

Feedback and suggestions are welcome. I can be reached by email, or on Twitter or GitHub.

Pingmon, a simple network monitoring tool

I pushed pingmon to GitHub as of a couple days ago. Pingmon is a simple connectivity and DNS monitoring tool written in Bourne-compatible shell, and should work with any UNIX-compatible OS (see the README for details). It is loosely inspired by NOCOL and SNIPS, which I've been using since the late 90s. Pingmon features color-coded output and is ideal for home use.

Feedback

Feedback and suggestions are welcome. I can be reached by email, or on Twitter or GitHub.

Hacker News Feeds, Rolling Time Windows, and Diamonds In the Rough

How many articles have you read this week advising you to stop talking about being an entrepreneur and start doing it? If the answer is "too many," you suffer from two problems. The first is covered in the aforementioned articles. The second is that you're spending too much time reading Hacker News. I can't do much about the first, but I'll give the second a shot.

This isn't the first post to tackle the flood of stories and discussions coming out of Hacker News. Jeff Miller covered it well a few months back in A Cure for Hacker News Overload, and there are a slew of alternative interfaces and visualizations on HNResources.

Rolling Time Windows

There is a lot of great work on HNResources. I've tried each and every visualization, and always hit one or two things that weren't exactly what I wanted. For the point threshold-based visualizations, I constantly found myself jumping between feeds. Week days are generally the busiest, and 100-point feeds worked well for that. On weekends, I'd need to jump down to the 50-point feeds to get a good amount of content, and on holidays, it was fewer points or bust. The Top X Stories formats resolved that issue and gave me a nice, normalized view, but it bugged me that rolling over the date mark would either lead to a bunch of low-score stories being posted, or there would be a big delay before stories were "good enough" to hit the list

My preferred approach: rolling time windows. Rather than building a list of the Top X stories by date, use windows for the past 24 hours, 24-48 hours, etc. These feeds cover what I've found to be a good amount of articles:

HN Top 10 Stories
HN Top 15 Stories
HN Top 20 Stories
HN Top 25 Stories

Since a lot of people visit Hacker News for the discussions, I've included a couple of those, as well:

HN Top 10 Discussions
HN Top 15 Discussions
HN Top 20 Discussions

Diamonds In the Rough

If you are someone afflicted with Hacker News addiction, you've probably heard more than one person say that the community and the discussions are the reason they visit. The problem with many of these techniques for distilling HN content is that you miss out on the smaller, intimate discussions that a tight-knit community provides. To address this, I've built another set of feeds that tries to capture this diamond-in-the-rough feeling. Each feed is comprised of threads that are discussion-heavy, but pass certain filters for minimum and maximum points and comments.

HN Top 5 Underrateds
HN Top 10 Underrateds
HN Top 15 Underrateds

Wait a minute, aren't these going to be a bunch of vi vs. Emacs threads, that result in tons of discussion without providing any significant value?

It's possible, but rare. To hit the Underrateds lists, threads need to reach the front page of Hacker News, have a minimum number of points (which vi vs Emacs rarely do), and a good discussion-to-point ratio.

Is it worth following?

I think so. After using it for a few weeks, I've caught a bunch of interesting content that I would otherwise have missed: Sparkmuse (HN), Tell me why I suck (HN), Ask HN: How would you monetize 360K people?, and Ask HN: Dropping out of college and moving to SF.

Details

Feeds are generated using Python 3, SQLite, EC2, and Google FeedBurner. Code is available on GitHub.

Feedback

Feedback and suggestions are welcome. I can be reached by email, or on Twitter or GitHub.