LISP stands for List Processing and it was originally designed by John McCarthy around 1958. It was the first language with a garbage collector making it the first truly high level language assuming you don’t consider Fortran a high level language. Here is Dr. McCarthy’s seminal paper and for a much better intro than I can give please see here.
Time passed and along came a man named Rich Hickey. Making a long story short, Rich was working in a variety of languages such as C++, Java, and C# when he did a project in Common Lisp and was hooked. There are many YouTube videos and documents that Rich has produced but simple made easy is one I found very compelling. If you enjoy that video, don’t stop there; Rich has many interesting, profound, and sometimes provocative things to add to the conversation. For more about his reasoning behind Clojure, please check out his rationale.
My perspective is closer to Uncle Bob’s.
To address the parenthesis, we need to talk about homoiconicity. LISPs are part of a subset of languages that build themselves out of their own data structures so that when you type symbols into a file or repl, you get back a data structure of the language itself. This means that several parts of the programmers toolbox can be simpler and you can load a file as data, transform it, and then execute some portion of it all without leaving the language. This isn’t something you will really need to understand today, but the point is that the look and structure of the language is a sweet spot to make it more of middle ground between what a human and a machine can understand.
The fallout from having a language that is both a language and a data structure is that you can extend the language without needing to change the compiler. For example, the very first standardized ‘object oriented programming’ system was CLOS, or Common Lisp Object System. This was implemented on top of Common Lisp with no updates to the compiler whatsoever. In Clojure, we have been able to add things like an async library or software transactional memory without changing the compiler itself because we can extend or change the language quite substantially at compile time.
Clojure is a deeply functional language with pragmatic pathways built for mutation. All of the basic data structures of Clojure are immutable. Learning to program in a functional manner will mean learning things like
reduce and basically re-wiring how you think about problems. I believe it is this re-wiring that is most beneficial in the long term regardless of whether you become some functional programming God or just dabble for a while.
Many systems, regardless of language, are designed in a functional manner because properties like idempotency and referential transparency make it easier to reason about code that you didn’t write. That being said, Clojure doesn’t force you to write functional code all the time as it is mainly a pragmatic language. You can easily write mutable code if you like.
No talk about Clojure would be complete without giving major credit to its excellent REPL support. One important aspect of the Clojure REPL is that you can see all of complex nested datastructures easily without needing to write
__str__ methods. Because of this visibility advantage a common way to use Clojure is to model your problem as a transformation from datastructure to datastructure, testing each stage of this transformation in the repl and just letting the REPL printing show you the next move. Programming with data is often just easier than programming with objects and debugging data transformations is far, far easier than debugging a complex object graph.
There are in fact many resources to learn Clojure and here are some the community recommends:
- Clojure for the Brave and True
- Clojure for Data Science
- Functional Programming in Scala and Clojure
Practicalli has a page devoted purely to resources on learning Clojure at whatever level you are so if these recommendations do not speak to you please review their books page.
There are many more IDEs available than listed here; these ones are just very popular:
One thing to be sure of, regardless of IDE, is to use some form of structural editing. All the better IDEs have it; all the IDEs listed here have it, and I personally really struggle without it. When I have a form of structure editing, however, I can move code around much faster than I can in Java, Python, or C++. This is another benefit of the homoiconicity that we spoke earlier in that we can transform the program easily because it is just a data structure and this includes editor level transformations and analysis.
Clojure is an amazing language. It is really rewarding on a personal level because it is tailored towards extremely high individual and small group productivity. But this power comes with some caveats and one of those is that learning Clojure takes time and patience. The community is here to support you, however, so check us out: