A cup of the good stuff

First, here's what a well prepared cup of my favorite beverage looks like:

That's a small cup, and there's probably a little less than 2 oz. in it. For some reason, in the United States, if you order an espresso, the "specialty coffee" chains think that you want a lot of it -- the bigger, the better, right? -- and end up over-extracting the coffee. That's why you almost have to make espresso at home if you want it done right.

Take a look at the cup. It's ugly. It's no collector's edition. But it is thick and heavy and holds the heat. (I also suspect that, if my espresso drinking were interrupted by a brutal criminal having the intent to do me harm, and I was forced to make a hasty defense, the cup could easily become a missile that would drop the bandit like a greased anvil. Something to keep in mind if you're in the market for espresso cups and it comes down to a toss up between these babies and some lightweight, designer-bait cups from Illy.)

Fedora Linux makes a mighty fine desktop

So far, I'm loving Linux on the desktop. When I switched my main workstation from Win2k, I was worried about losing support for add-ons, such as my HP OfficeJet G95 and my Palm Pilot. Happily, I have found my worries to have been unmerited.

The HP OfficeJet Linux driver, which again is already included with FC1, does a darn fine job. Setup was a breeze, and now I'm printing and scanning as usual.

My Palm Pilot syncs with J-Pilot, conveniently part of Fedora Core 1. I just hooked it up and it synced on the first try.

I'm still debating whether to get VMware in order to run my few Win32-only apps.

The PFC wait begins

Now that the submission period for PFC: Tragic has ended, I'll share my thoughts. I was delighted to see entries like edwin's network flow–based approach. Most of us PFC old timers fell too quickly into traditional searches and missed the opportunity to fully explore the interesting attributes of jacob's well-chosen challenge task. Woe unto me that I ever read Korf's survey on searches!

I love edwin's entry. The hardest part of solving problems is not applying the chosen means of solution but rather choosing the best means in the first place. Edwin chose a better means than I and the rest of the searchers did. He also gets bonus points for choosing a means that could be implemented in terms of readily-available libraries. It's a great entry all around. Props to edwin.

I thought it was interesting that celeriac use linear programming, in particular the simplex algorithm, in his solution. I actually gave a talk on the simplex algorithm about ten years ago, but like a fool I didn't see the now-obvious Mr. T connection. Doubtless my talk suffered for it. Props to celeriac for making the simplex algorithm part of the A-Team. (See poll.)

That's why I love PFC. It's fascinating to see what approaches people will choose. I hope that jacob has the time to classify the entries based on their means of solution and to provide commentary on each class. My hunch is that the final ranking will be something like this:

• network flow
• linear programming (less efficient)
• optimized search
• brute-force search
• things that don't really work

Jacob has a tricky task ahead. I suspect that some of the search-based solutions will actually perform very well for real-world decks. So the question becomes one of how to balance theoretical and real-world algorithmic efficiency. I'm glad it's not my problem. (Have fun, jacob! Tee hee hee.)