Installing Python Audio Tools

On Linux Mint, make sure the following are installed:

python-dev
libcdio-dev
libcdio-paranoia-dev

Then

sudo make install
Also, for reading mp3 files: sudo apt-get install mpg123

Using Java Aspects to Support Monitoring

I am running a Java/Spring application in Heroku's Platform as a Service using New Relic as the monitoring and performance analysis tool. Normally, New Relic makes it easy to see what web transactions are doing because they are usually named by their URL.

However, for multiple reasons (framework, coding conventions on our part), the transactions all lumped under a generic name (e.g., /servletName), adding a layer of indirection in identifying problematic pages or REST API calls at a glance. No problem - New Relic supports multiple ways to set a name for a transaction, including via a request attribute. The challenge is ensuring that all methods that we want to track get a meaningful name. The ugly solution would be to mine the code and sprinkle the logic everywhere. Less ugly would be finding a way to do it via a base class or interceptors.

This turns out to be a perfect use case for aspects - setting a transaction name for a third party monitoring tool is a cross-cutting concern that can and should be decoupled from the mainline logic of the system.

The solution requires three pieces -

• The aspect to set the transaction name
• A marker annotation used to identify the join point
• Annotating the relevant controllers (or base classes)

In addition, since we wanted as close to a 1:1 relationship between transaction and controller entry point, we specified two conditions for the join point - our marker annotation, and the spring RequestMapping annotation. Where appropriate, the annotation was added to a common base class and flagged @Inherited.

The aspect: The annotation: After that, getting the correct name to appear on for our transactions was a matter of putting @NamedTransaction on the common base class for our controllers. Now, the New Relic transactions show up as method calls, leading us straight to the problematic transactions. E.g., @ProfileController.getContacts(...)

Dork Alert

I heard Led Zeppelin's "Ramble On" the other day, and found myself obsessing over the Tolkien references in the lyrics. In a moment of grim realization, I realized I couldn't figure out which was worse:

• Spending mental energy trying to derive a metaphorical interpretation ("Oh man! The girl is the RING!")
• Fretting that the lyrics don't align to the LOTR canon to begin with

Help?

From Internet Debunking to Wobbly Knees

I received yet another one of those infinitely forwarded emails from a family member that gets my skeptic gears a-churning. This particular email purported to show some harrowing ascent up sheer granite cliffs to a restaurant or tea house at the top of a mountain in China. This is combined with a promise that if one is brave enough to make the trek, then the food is free. It's essentially an email version of this (go ahead and look... I'll wait), with some added flair and the bit about free food.

Hm. That just seems a little too out there to be true, so I quickly set out to debunk the whole thing. Oh, how fortune taunts me!

From what I can find, the harrowing ascent is real, although the destination is not a restaurant, but a temple on a peak of Huashan Mountain, one of five holy mountains in China. What's more, you can find videos of people documenting their turn on the "plank walk" - the part of the hike where you traverse a stretch of cliff on rickety boards. Take a look:

Oh my. Here's the link if you can't see the embedded video: http://www.youtube.com/watch?v=4nq0vtU-LVc

And of course, here's where the interwebz works its dark magic . YouTube helpfully suggested related videos that, in the spirit of "oh-my-god-i-can't-look-away", I had to watch. I kept returning to two knee-wobbly favorites. The first is a trip along "El Camino del Rey" - a 100-year old gangway built for workers at a hydroelectric plant in Spain:

Again, the link if you can't see the embed: http://www.youtube.com/watch?v=y1Nd1qtk1Go

The other is a classic of a tower worker getting to the top of a 1700+ foot transmission tower. Unfortunately, I can't link to it because it seems to get taken down every few days. Try googling for "climbing tall tower" and are likely to stumble across it, though. What I love about this one is that it just keeps on going. Every time you think he's about to top out, there's some crux - a ledge to mount, an ever-narrow ladder to mount.

Yet again, reality is way more fun than the internet rumor mill!

How Googley

Useless trivia: Google owns a domain called 1e100.net. This odd name is symbolic for... wait for it... 'googol', which is, of course, 1 x 10¹⁰⁰, or 1e100 in scientific notation. Link.

Did I say useless? Turns out this is actually useful information to have, should you have kids and hold the position "Household CIO."

Why, you might ask? I was setting up parental controls for my kids' accounts on a new computer and terribly befuddled when my daughter couldn't connect to her GMail account. Other sites worked fine... the main Google site, Webkinz, etc. Turns out it's a combination of a few different things.

Firstly, OS X website/content controls work in one of three modes: unrestricted access to websites, access only to allowed websites, and "best attempt" filtering, dynamically allowing or denying access based on the website content. I had my daughter's account set to the latter, hopefully striking a good balance between access and protection - such are the aims of the Household CIO.

Secondly, under that setting, OS X tries to examine the content of a web site to determine whether to block it (looking for naughty words and such). GMail, however, forces the connection to SSL. When accessing sites with SSL, the content is encrypted such that OS X can't read it. Thus, it refuses the connection. The solution is normally to add the site in question to the list of allowed sites. Ok, fine... add a few https URLs (www.google.com, gmail.google.com). Good to go.

But wait! That doesn't fix the problem. Turns out all Google applications communicate extensively with this mysterious 1e100.net; haven't figured out exactly how it's used yet (working on it), but adding 1e100.net to the "Allow" list enables gmail to work.

Frustration gives way to enlightenment. Or at least mild amusement at a next-to-useless factoid.

How high are buildings in DC?

I have this annoying habit of replying to chain emails I receive from friends and family - notably to debunk whatever claims are in the email. I received one such email recently that contained a typical myth about the skyline of Washington, DC, notably that "no building may be taller than the Washington Monument". I've heard variations that involve the Capitol or the White House, but they all have the same general idea.

Now, technically it is a true statement - in the same sense that if a < b and b < c, then a < c. In this case, 'a' = the height of a building, and 'c' = the Washington Monument. The reality to the myth is that 'b' is set by an Act of Congress = the Heights of Buildings Act of 1910, itself a refinement of a previous act from 1899. In a nutshell, it sets the limit on the height of a building at 20 feet greater than the width of the right-of-way of the street the building fronts. A building on a street that is 90 feet wide can thus be no higher than 110 feet. There are, of course, refinements (commercial vs. residential, special treatment for specific blocks of Pennsylvania Avenue) and exceptions (the National Cathedral, the Basilica of the National Shrine of the Immaculate Conception). So, as with many things, there is no absolute.

See: http://www.ncpc.gov/ncpc/Main(T2)/About_Us(tr2)/About_Us(tr3)/HeightofBldgs1910.pdf

See: http://en.wikipedia.org/wiki/Heights_of_Buildings_Act_of_1910

Does this look all shiny?

This is posted from Chrome OS, the open source operating system under development from Google. It's basically a boot directly into the Chrome browser on a stripped down Linux kernel (see http://www.chromium.org/ for more information). I got a virtual machine disk image from gdgt.com and have it running under VMWare Fusion. So, performance is sketchy, but it seems to work as well as the Chrome browser itself, which is to say well enough. Minor note getting up and running - not sure if this was real or imagined, but I had to use "Bridge" networking in the VM rather than "NAT".

While originally skeptical of yet another OS, I can see this being an attractive option for people who have embraced the "online" lifestyle, have put most of their data in the cloud, and have multiple machines - real boxes for "work" running standard desktop software and lightweight boxes for the road running Chrome. You boot up, sign in with your Google account, and have links to all your Google apps (Mail, Calendar, Documents, Reader, Picasa) available on your "desktop". The one piece that could be a killer app is missing - Google Wave integration. The big idea there would be about making the waves an integrated part of the desktop experience. Everyone I've spoken with agrees that Wave will be successful based on the availability/ubiquity of clients. OS-level integration seems to hit that mark very well, especially if it can integrate activity across all the Google applications.

Another interesting question is how Chrome OS integrates with Google enterprise "apps" accounts (like Google Apps for Education). Assuming that's possible, add support to Chrome for things like remote desktop (RDC, VNC, ssh terminals, etc) and VPN for remote work on more powerful workstations/servers, and you make a compelling case that the netbooks for which Chrome is likely to be targeted can be more than over-sized smart phones, and instead can be viable extensions of many work environments (home, office, or other).

Overall, a fascinating development. My earlier skepticism is waning and I'm interested to see how it evolves.