[rcurry]

I actually witnessed a funny incident, similar to this, when I worked at a National Laboratory some years back. A team I was working with had a big project that formed the core of a larger system, so they created a directory named "core" and spent the next N months happily churning out code. Meanwhile, the system administrator had a set of backup scripts that he'd configured to ignore core dump files (yeah, you can guess where this is going), but the scripts that he wrote didn't differentiate between files or directories that had "core" in their name, so nothing got backed up for almost a year. Over the Christmas break, the system administrator decided to install a new version of Solaris, and wiped all the drives as well. I came in to work after the break, and everyone was freaking out because all their code was gone. They managed to recover some of it, because some developers had separate copies of parts of the system, but it was still a disaster.

Every once in a while, at one job or another, someone will suggest we name a source code folder "core" and I get to relate that story all over again.

[wildmusings]

>I came in to work after the break, and everyone was freaking out because all their code was gone.

What happened next? This cliffhanger is killing me.

[rcurry]

I honestly don't know how they recovered from it all. I think they were able to recover parts of it from some people's workstations, but still had to rewrite an awful lot of their code.

[mlvljr]

Once again proves, you have to A) look at what your scripts are actually doing and B) test them manually at least couple times somehow, otherwise this is plain unprofessional (in one of the many ways sadly and funnily possible).