I remember one cathartic event probably in 1988. I was working on a Sun 3-50 when I noticed a slowdown. I did a ps (processes status command) and noticed two things. A colleague was connected to the 3-50. And two consecutive ps gave me very different pids. I asked the colleague what he was doing. He was using sh or csh to do arithmetic. Each arithmetic operation was costing a fork.

I correlated that with the difficulty of building scripts with pipes, double quoted interpolation, bactik interpolation. I realized that the "Unix mantra combine small tools to make bigger ones" did not scale or necessited competences beyond mines. Also I had to remember many small languages and the command line switch for many utlities. Many of them supported some set of regexps but not the same subset and not always the same syntax. What a mess.

I was a C programmer but I liked the shell concept, Building thinks incrementally and testing at the command line was good. I searched mostly self-sufficient "shellish" tool and I found perl. The manual was not too big (this was in 1988) and I soon realized that was very the tool I was longing for. So I tried a statement with a regexp (probably incorrect) and the interpretor crashed. What a disappointment. I checked again a few month later and I had no problem. In fact you can see my first post dated 17-12-89 to comp.lang.perl immediately corrected by "Tom C.". I am talking about my first significant program at stefp (scroll down and read at the right of the HP25 picture -- me too lazy create a link :)

-- stefp