I've established that I was hired as a "button pusher tester", but at least the work was interesting. The thing that's funny is that ten years ago, I really didn't know much about testing. I was a "tester", but what I was really good at was figuring out how things work (which is really only a single trait of being a good tester). I also had a reasonable level of programming experience that I wasn't supposed to be using.
Anyway...my job was to test networking on the Japanese version of win95. These days, I know a bit of Japanese, but at the time, I didn't know anything (nor is that how I learned). Any tester worth anything should be able to test localized software without knowing the language. They can't test localization, but they certainly should be able to get around even if at first it requires comparing with an english build side by side. Anyone who has done this won't bat an eye when I say the side by side comparisons only lasted a few days before I could get around (remember, I only had to get through installation and networking component related strings, so there wasn't a lot of navigation to learn).
With win95, compatibility was king. The most time consuming part of my job was installation testing. Win95 shipped with fifteen or so japanese specific network cards - about 5 on the PCAT platform, and 10 on the PC98 platform (not nearly enough space here to explain the PC98 platform). At the time, no one considered equivalence class testing, or any sort of diagonalization of the test matrix, so for each of the 15 network cards, I had to make sure it worked on both NT and Netware networks on a clean install, as well as with upgrades over win31, lanman, and netware dos clients. I think netware also supported 2 16 bit clients, but I can't quite remember. On the upgrades, I had to verify that the IO and IRQ (remember those?) settings were preserved. For clean installs, I had to verify that every supported IO / IRQ range worked - which meant multiple installs. Luckily, I only did this testing once a milestone. Believe it or not, it was fun (probably only because it only needed to be done every 3-4 months. On days when I knew I had to do upgrade testing, I would come in late, take care of any pressing details, then go to lunch early. The build came our around noon, and I would blast through the entire installation matrix in the next 14 hours or so. I was so focused on keeping the installations optimized across all of the computers I was working with that the time flew by. I almost always found a bug or two, and these bugs were always fixed (proof that compat was king). One of the bugs I found was in the inf file. Remember that these network cards were Japanese specific, and the turnaround time from the OEM was usually two weeks or more. In this case, we didn't have time, and I knew (and had tested) the fix. So they let me fix it - my first shipping code!
I have one follow up post on the cast of characters I had to work with, then I'll leave nostalgia behind me.