Irreal

A few days ago I wrote about the excellent video of David Brailsford and Brian Kernighan discussing AWK and its history. In the video, Kernighan mentions that he’s been working on enabling Unicode in the One True AWK. Here’s a pull request from Kernighan showing that he’s mostly accomplished that goal.

At one level, it’s easy to believe that it’s basically a trivial change but as AWK demonstrates it’s not always so easy. Probably the hardest thing is fixing AWK’s regex parser to accept and deal with Unicode. But even simple things like calculating the length of strings can be a problem.

When AWK was first developed—and long afterwards—ASCII was sufficient. These days, it’s a real imposition to deal with an app that doesn’t support Unicode. Kernighan’s porting AWK to support it will ensure that AWK will continue to be a useful tool not only for English people speaking people but for those who speak languages with non-ASCII characters as well.

If you’re a young engineer, the idea of open source and having access to the source code to your tools seems unexceptional. That’s just the way it is. But AWK comes from a time when that wasn’t true. It’s great to see the original AWK still available and still under development. AWK and its developers are a national treasure that we should all be thankful for.

Recently, I wrote about a Janet Jackson song that could cause laptops to crash. That turned out to involve frequencies from the song that resonated with a critical frequency in the disk subsystem and was solved simply by installing a filter to damp out the offending frequency.

Right after I published that story, I stumbled upon a video from 14 years ago of someone shouting at a disk array. It didn’t cause a crash but monitoring software clearly showed that disk latency went up when the disk was shouted at.

The engineers doing the demonstration explained this as having to do with vibration so perhaps it’s different from the Janet Jackson menace but it’s surprising how unexpected things can affect disk performance.

Raymond Chen occasionally posts interesting stories from his (long) time at Microsoft. His latest offering tells the story of how Janet Jackson used to have the power to crash laptops. It turned out that playing Jackson’s Rhythm Nation on certain laptops would cause them to crash. A little experimentation showed that playing the music on one laptop could even cause another nearby laptop that wasn’t playing the music to crash.

I’ll let you read Chen’s post to see what was happening and how they fixed it but the interesting thing is that Chen speculates that the fix may still be in place even though the hardware involved is no longer used. It was one of those things were the fix was installed with instructions that it should not be removed and years later no one knew why it was there but were afraid to remove it.

It’s a real problem. Sometimes, like in this case, the fix is no longer doing anything useful but sometimes removing it without a thorough understanding of what it was doing could lead to disaster. Things like this are what make our industry so endlessly engaging.

Computerphile has a another wonderful discussion between David Brailsford and Brian Kernighan. We are quickly reaching the time when all the original Unix people will be gone (Kernighan is 79 or 80) so these chats are our last chance to get an oral history of what it was like in the beginning.

This particular chat is about AWK. I thought that by now everyone knew that the K in AWK stands for Kernighan but judging from the comments, apparently not. AWK dates back to the 1970s and is still maintained—even the original AWK—as well as the GNU version GAWK. It’s my favorite scripting language and tremendously powerful for problems in its domain.

One of the things Kernighan revealed in the video is that he’s recently spent some time in making (the original) AWK work with Unicode and that his summer vacation project is to update the AWK Book, which if you follow the video link you’ll learn is from 1988. It’s still available but at a outrageous price so a new version would be very welcome, especially to younger engineers who may not have access to the original.

I always enjoy these Brailsford/Kernighan chats and inevitably come away from them knowing something I didn’t know before. In an age where many people in the field don’t know that Kernighan is the K in AWK or even that he’s the K in K&R, these videos become more important than ever.

For some reason there was a recent pointer to this 6 year post by Matthew Jones on some of the fundamental laws of software development. Most of them will be familiar to Irreal readers but it’s nice to see them listed along with their explanations in one place.

Jones lists 15 laws or principals. They are:

1. Occam’s Razor
2. Hanlon’s Razor
3. The Pareto Principle
4. Dunning-Kruger Effect
5. Linus’s Law
6. Robustness Principle
7. Eagleson’s Law
8. Peter Principle
9. Dilbert Principle
10. Hofstadter’s Law
11. The 90-90 Rule
12. Parkinson’s Law
13. Sayre’s Law
14. Parkinson’s Law of Triviality
15. Law of Argumentative Comprehension

Some of these, like Hanlon’s Razor and the Dilbert Principal, are tongue-in-cheek while others, like The Pareto Principal and the Dunning-Kruger Effect, are serious, a précis of actual research.

Oddly, the most famous law of all, Murphy’s Law, doesn’t make an appearance. As every developer knows, it is always with us and operative. Regardless, the list is amusing and worth taking a look at if you’re searching for a momentary diversion.