That was the way it was supposed to happen with Computing Unplugged. The reality was a bit different. The first part was true, in that we did work for months to put this magazine together and we did do the job quite well. Unfortunately, fate (in the guise of some mean-spirited open source software) got in the way.

About three weeks ago, we published our first issue and sent out our first update mailing for Computing Unplugged. I was so darn proud. I went to bed that night excited and feeling good.

The next morning, I woke up to HELLLLLLLL!

There's a very long story that involves two weeks of trying to understand why the server melted down, but the end result was that PHP, the programming language we use for some of our systems, had a problem with doing something called an "include" (for the techies in the house, it would allocate handles, and never return them) and when it hit the level of traffic that the new publication added, it took the Web server down.

Unfortunately, the Web server was based on Linux, and for some reason (the Linux guys tell us we're not smart enough to understand), when the Linux box went down, the entire hard disk infrastructure would corrupt terribly, and we'd have to rebuild the thing. We were generally rebuilding the Linux server every three to five weeks, and the morning after we launched Computing Unplugged, everything was toast. We couldn't even get the box to boot, we couldn't get the disks to restore, it was a total mess.

This time, however, after rebuilding the system, the added load torched it again, triggering the PHP include bug, and nuking the box. We, uh, couldn't get it up.

Unfortunately, it took quite a long time to track down the bug to PHP and include, and to realize the problem was most pronounced under Linux. And so, we've been spending the past few weeks feeling a bit like NTSB (National Transportation Safety Board) inspectors at a particularly mysterious crash.

Finally, about a week ago, we were able to confirm the problem and figure out a work-around. The good news was we'd found a work-around. The bad news was that we had to re-program a large chunk of our journal system to work with the work-around and then re-render every single page from every single issue of every magazine we'd ever published.