Here is a command that really opened my eyes to how cool sed, grep and piping are:

ls -R | grep ":$" | sed -e 's/:$//' -e 's/[^-][^\/]*\//--/g' -e 's/^/ /' -e 's/-/|/'

courtesy of this site: http://www.centerkey.com/tree/

This ties in nicely to the threads about the necessity of comments. :)

As a bonus, most (all?) of the sed commands are also usable in vi's ex mode (the : prompt).

How much value is there in learning sed when you go to languages like Perl or Python for such tasks?

Perl works as a hyper-sed, but I think sed and Python are apples and oranges.

Python gives me access to a complete regular expressions engine just as Perl does.

I know. So does OCaml. Perl is a lot closer in character to sed and awk than Python is, though. That's all. I really dislike Perl, but I'm quite happy with using Python for general scripting and using sed for quick regex piping tasks. If I used Perl instead of Python, I probably wouldn't bother using sed.

Perl has a lot of syntactic sugar for quick-and-dirty regex stuff, derived directly from sed and awk, while Python instead has sugar for list slicing, dictionaries, etc., but uses function syntax (e.g., re.search(pattern, string)) for regexes. I would rather have great syntax for lists and dicts, personally.