Over in the Facebook AP Computer Science Teachers group someone asked for thoughts on covering BASH scripting as a post AP topic.
A number of us made suggestions. I linked to this old blog post.
One group member said she asked around for similar suggestions and the response she got was "vi and awk." I wanted to jokingly respond "and after they suggested that they got into their time machine and went back to the 70's."
In all seriousness though, I think that suggesting specific tools or commands is off base.
The important thing to know about Vi is how to get out of it but it isn't really a tool in the scripting sense. I do think students should spend a good amount of time learning a powerful editor and should try bot Emacs (my choice) and Vim but that's another story.
I also use AWK but as it's a programming language in it's own right, I'm not sure if I'd introduce it right off the bat.
There are a number of important ideas kids can take away from learning some Linux (or other Unix flavor):
- There's something out there besides Windows and MacOS
- All about free software
- The Unix Philosophy
That last one is the biggie and more specifically, there's a huge upside in teaching kids the value of "OS as Toolset" where they can compose the many tools that comprise the Linux experience to get things done.
I gave an example of that in the post I previously linked to.
For the teacher, that means wrapping your head around that way of working. Living in the shell and using pipes to connect program to progarm to program.
I'd recommend getting into a time machine ourselves and taking a look at:
It's dated but it's really a great book on getting into the Unix way of doing things, particularly the chapter about filters. It also has one of the best and clearest introductions to writing a compiler in the chapter on program development.
As I said, it is dated - shells are much easier to use and much more robust, there are many more tools now, and they've evolved but it's really a must read book.
In terms of tools, I get a lot of mileage out of:
|cat||catenate or display a file|
|tr||Translate characters||tr A-Z a-z||convert upper to lower case|
|sed||Stream editor||sed "s/a/b/g"||Replace all a with b|
|wc||word count||counts words lines and chars|
A nice simple thing you can do with these is clean data. Let's say you want to do some analytics on a book from Project Gutenberg. You might want to convert all non letters to spaces, and all letters to lower case:
cat book.txt | sed "s/[^a-zA-Z ]/ /g | tr A-Z a-z"
That sends book.txt into sed which uses a regular expression to convert no space and letters to spaces. The tr command converts all upper case letters to lower case.
If you want one word per line, add:
| sed "s/\n/g"
and maybe get rid of blank lines:
| sed "/^$/d"
We can now count the number of words in the file using *wc or even get counts of all the words:
| sort | uniq -c | sort -n
sort will sort all the lines, uniq -c will compress the lines that are adjacent and the same and give you a count and then sort -n will sort the results numerically.
I wrote another post a while ago about using the shell to detect who responded on a Google form. It looks like it didn't convert when I moved to my current blogging platform - I'll repost that shortly.