The only time I made serious use of big pipelines was when I was trying to diagnose some bug in the production system using the log files. At that time, I would

The only time I made serious use of big pipelines was when I was trying to diagnose some bug in the production system using the log files. At that time, I would make a pipeline with lots of `head`, `tail`, `grep`, etc. I would reorder the filters, change the file it is running on, sometimes run it on multiple files. Using `cat` to feed the initial input was way better than trying to avoid UUoC. Concatenating files became possible, reordering became simple. It was just overall a much better experience with `cat`. In my daily life if I am writing shell pipelines, I still use cat because it just flows better in my brain. Step 1, take the input; step 2, apply filter; step 3, apply other filters; step 4, print it out. When I don't even know which filter I am going to be using, it is easier to write the cat and glob the files while my brain catches up on whether I want grep or sed or something else.