If you mean the repetitive spaghetti of pseudo text processing in init.d, then yes I've seen it. Everything tries to do the same stuff in a different way, there's lack of one framework, error checking is present only in random places and any script can hang because there's no real timeout handling. They're also different in almost every distribution and some are incompatible between sh/bash and very few people can tell the difference. The whole idea is one level too low and we're finally growing up to use something simpler.
That's one of my "favourite" examples of using scripting where it's not needed. Have a look at what's actually keeping critical systems working -> inittab, supervisord, etc. Lately upstart also migrated from scripts to something more declarative.
Edit: forgot the bit where everyone "parses" the output of commands like `ifconfig` and pretends they will never change, or will never fail, or will return the information in the same format. How many scripts relied on grepping ifconfig output for "inet" before "inet6" broke the startup?
Ultimately the point was that shell scripting shouldn't be used for "serious" programming. Yet it has been used for decades to accomplish and automate some of the most important tasks on UNIX/Linux systems. I happen to know and work in several "serious" programming languages, and yet I can knock out a functional, reliable, and understandable shell script in a fraction of the time it would take many people to do the same in X.
You mean the scripts that start thousands of processes before my computer is even ready to use? Especially things like sed, awk and perl just to do some text processing?
