I commute from Margao to Bambolim every day. Normally a 30-minute drive. Today it took me an hour and forty minutes. There’s road construction near Verna that started four days ago. Four days of digging with no visible progress. And they’ve decided the best time to work is during peak hours – 9:30 am to late evening. Exactly when everyone is trying to get to or from work. I was late for an important meeting today. Had to apologise to my boss and to clients. All because someone thought it was a good idea to block a major route during rush hour. ...

The goal of encryption isn’t to create something uncrackable. The only truly unhackable computer is one that’s turned off, unplugged, and locked in a vault – and even then, someone could just carry the vault away. Encryption is about making it difficult enough that attackers give up and move on to easier targets. It’s a cost calculation, not a guarantee.

Everyone tells you to go to college. It’s the default path. Finish school, get a degree, find a job. But is it actually worth it? The tuition costs are rising, graduates are struggling to find work in their field, and student debt is becoming a genuine burden. For some people, college is absolutely the right choice. The networking, the structured learning, the credential – it all adds up. For others, it’s four years and a mountain of debt for a piece of paper that doesn’t open any doors. ...

Sat at CCD today with friends and we ended up looking up people on Lemonlaw. It only took five minutes. I know that’s not the point of the thing, but it felt wrong how easy it was.

A friend mentioned recently that we’re going through too many changes. I didn’t really know what to make of it at the time, but it stuck with me. Later that day I did something daft — I reverted my Facebook profile picture to an old one. I knew it was pointless as soon as I did it. Changing a profile picture doesn’t anchor anything. But I think I was reaching for something familiar, something that hadn’t changed, even if it was just an image on a screen. ...

Google announced a new programming language called Go on November 10th. It’s from Robert Griesemer, Rob Pike, and Ken Thompson. That last name should tell you something — Ken Thompson co-designed Unix and invented the C programming language. The pitch was simple: a compiled language that feels as quick to write as a scripting language, with built-in concurrency and garbage collection. It’s aimed at the kind of problems that C was always used for, but without the pain of manual memory management and with better support for multi-core processors. ...

After building and maintaining CMS sites for a while, here are the things I’ve learned the hard way. Don’t run your site from the root directory. Put the CMS in a subdirectory and forward requests there with .htaccess or whatever your server supports. It keeps the CMS files out of the way and makes it harder for someone to guess where things are. A small thing, but it helps. Be careful about advertising what CMS you’re running. ...

Mandriva Linux 2010 is launching on November 3rd. Highlights from the announcement: KDE 4.3.1, GNOME 2.28, kernel 2.6.31.1, and Xorg 1.6.4 RC 1 with updated Intel graphics drivers. There’s also a Moblin 2 spin for netbooks. I’ve used Mandriva off and on over the years. The Control Centre has always been one of the better configuration tools in the Linux world. Worth keeping an eye on to see how this release shapes up.

This is the process I follow on projects. It’s not fancy, but it works for me. Defining the requirements. Write down what needs to be built. Be specific. Ambiguous requirements are the root cause of most problems later on. Approval. Get sign-off on the requirements before anything else. If the requirements change later, that’s fine, but do it deliberately, not by accident. Template designs. Work out the structure and layout before writing code. For web projects this means mockups and wireframes. For other projects it means architecture diagrams and data models. Template approval. Get the designs signed off too. It’s cheaper to change a design than to change code. Coding. Build it. Internal release. Get it in front of the team first. Internal testing catches the obvious stuff before anyone external sees it. Testing. Proper testing, not just “I clicked through it once.” Automated tests where possible, manual where necessary. Alpha release. Early access for a small group. Expect bugs. The goal is to find them in a controlled environment. Beta release. Wider release. More users, more edge cases. This is where you learn what people actually do with the software versus what you thought they’d do. Project goes live. Production. The real world. The key thing is not to skip steps. I’ve seen projects that jump straight from requirements to coding and then spend three times as long fixing problems that a proper design phase would have caught.

I spent some time at Arsin Corporation, a company based in Hyderabad, India that specialises in enterprise software testing. Here’s a summary of what they do, mostly for my own reference. What they do Arsin provides automated testing services for large companies, with a particular focus on SAP. They’ve been around since 1995 and their clients are mostly Fortune 500 companies in pharmaceuticals, finance, healthcare, and technology. The idea is that when you’re running enterprise software, you can’t afford for it to break. Arsin builds the testing infrastructure that catches problems before they reach production. ...