4. Complex Systems
… are simply not
easy to understand
@jr0cket
Josh Smith - Tweet
Original Gif
5. The Complexity Iceberg
- @krisajenkins
● complexity is very
dangerous when hidden
● You can't know what a
function does for certain if it
has side effects
7. Pure Functions
The results of the function are purely determined by its initial output and its own code
- no external influence, a function only uses local values
- referential transparency (the function can be replaced by its value)
8. Impure Functions - side causes
The results of the function are purely determined by its initial output and its own code
- behaviour externally influenced and non-deterministic
9. Eliminating Side Effects
Functional programming is about eliminating side effects where you can,
controlling them where you can't - @krisajenkins
The features in Functional Programming come from a
desire to reduce side effects
11. Clojure / ClojureScript
A hosted language with simple interoperability with the host language
- (java.Util.Date.)
- (js/alert “Client side apps are easier in Clojure”)
12. Clojure - basic syntax for this talk
( ) ;; an empty list. The first element of a list is evaluated as a function call
(function-name data) ;; call a function with the data as its argument
(def name “data-or-value”) ;; assign (bind) a name to a data or legal value
:keyword-name ;; a keyword is a name that points to itself
;; Thread-first macro - chain function calls, passing the result of each call as the first
argument to the next function. The ,,, indicates where the resulting argument goes.
(-> (function-a “data”)
(function-b ,,,) ;; In Clojure commas , are whitespace
(function-c ,,, “data”))
14. List, Vector, Map & Set
Clojure’s built-in data structures are all immutable
- returning a new data structure when a function is applied
(list 1 2 3 4 5) ‘(“fish” “chips” 42)
(vec ‘(1 2 3 4)) [1 2 3 4]
{:key “value”} {:name “John” :skill “conferencing”}
(set ‘(1 2 3 4 4)) #{1 2 3 4}
15. Persistent Data Structures - shared memory
Each function creates a new vector
Memory space for values is shared
between each vector
16. Persistent Data Structures -
shared memory
By sharing memory you
can apply functions
over and over again
effectively
Values persist until
they are no longer
referenced
17. Concurrency is Easier
Concurrency is much easier to write and reason about because of
- Pure functions
- Immutability is encouraged by default
- Persistent Data Structures
- All values are immutable
- unless explicitly wrapped in an atom or ref
- state changes managed atomically (software transactional memory)
- core.async library allows you to write asynchronous code as easily as sequential
code
20. Sequence / List Comprehension
Iterating through a range of generated values to create a list of 2 value vectors
21. Immutability - local binding
Assignments made locally are immutable
- words is a local binding to the result of running the function upper-case on “Hello
World”
- letter->clack is a function that converts a character to a code
23. Lazy Evaluation
Clojure's lazy sequences
- returns a reference to a file and step through it one line at a time
- line-seq returns a lazy sequence of lines, so we can read files larger than
available memory, one line at a time
25. Recursion & Polymorphism
Process a collection of values by feeding the remaining elements back to the function
- the sum function is polymorphic, it has different behaviours that could be
evaluated depending on if passed 1 or 2 arguments
26. Recursion - tail call optimisation
Protect the heap space from blowing by using the recur function
Using recur in the last line is the same as calling sum, however the memory required
from the previous sum function call is over-written in memory. So only 1 memory slot
is used instead of 10 billion
27. Higher Order Functions
A function that takes one or more functions as arguments, or
a function that returns a function
29. Composing functions together
Example: current value of the Clojure project from the configuration file
- `slurp` in the project file, convert into a string and return the value at index 2
31. Functional Composition - map
Map the function inc over each number of the collection,
returns a new collection of numbers incremented by 1
32. Functional Composition - reduce
Reduce the numbers inside a collection to a single value by applying the + function,
returns a single value
33. Higher Order functions - partial, comp
Higher order functions can return a function
- The comp function composes other functions together
- The partial function allows you to lazily add another argument to a function
35. Safe State changes
Changing state safely by not changing it
● Persistent data structures
● Local bindings
Changing state safely by changing it atomically
● Software Transactional Memory (STM)
○ Gives an mechanism like an in-memory atomic database that manages mutable state changes
under the covers
● Atoms
● core.async
36. Concurrency syntax - atoms
An online card game has players that can join and have their winnings tracked
37. Concurrency syntax - atoms
The join-game function adds players to the atom by their name, but only up to 2
players
38. Concurrency syntax - refs for sync updates
The join-game-safely adds players to the ref and alters their account & game account
39. Putting it all together
Let's find all the most common words used in a popular Science Fiction novel
54. Over 20 Books on Clojure...
Where to start with Clojure will be different...
Example:
I typically suggested BraveClojure.com as a starting
point, however many people prefer LivingClojure or
ClojureScript Unraveled...
Help people understand the relevance of a book and if
it's the right thing for them at that time.
63. Clojurian Community in Person
Probably the most active language-specific
developer communities in London
64. Learning by teaching others
I really started thinking in Clojure when I started talking to & teaching others
- Coding dojos
- talks on Clojure (starting with the basics, showing the art of the possible)
- moving on to running conferences
- workshops at hack days