8.5

### The Style

In addition to following the design recipe, all code must adhere to the following style guidelines:

### Document Structure - General Exercises

1. Organize your program top-down. This means that when you write a solution that contains multiple functions, the primary function should come first, followed by helpers. The functions should be in order of where they appear in the primary function. For example, the following code is organized top-down:
 ; my-function : Number String -> Number ; Add double the string-length to twice the number cubed (check-expect (my-function 2 "hi") 20) (check-expect (my-function 3 "hello") 64) (define (my-function n s) (+ (double (cube-num n)) (double-length s))) ; double : Number -> Number ; Computes 2n (check-expect (double 4) 8) (define (double n) (* n 2)) ; cube-num : Number -> Number ; Produces the cube of this number (check-expect (cube-num 5) 125) (define (cube-num n) (expt n 3)) ; double-length : String -> Number ; Produces twice the length of this string (check-expect (double-length "goodbye") 14) (define (double-length s) (double (string-length s)))

Please note that many of the above helper functions were written solely to illustrate the top-down organization we expect.

2. Title your exercises. Above every exercise please note which exercise it is.

3. Separate data definitions. Data definitions (and their corresponding examples/templates) should be placed at the beginning of the relevant exercise. Data definitions do not need to be repeated if used in multiple exercises.

### Document Structure - Large Programs

1. Organize your program top-down, regardless of how you actually work through your wish list. The phrase "top down" means that project files consist of a data definition and a constant definition section, a main function, followed by sections for handler functions, and wrapped up by general utility functions (functions used by multiple handlers). Within these sections please use top-down organization as defined above.

The main function is the one that uses big-bang, read-file, write-file, and so on. A good purpose statememt for the main function explains how to use it. For example,
 ; Number -> Number ; (main n) runs Typaholic! with tick-rate n (define (main tick-speed) (... (big-bang WORLD-STATE-0 [on-tick update-world tick-speed] [to-draw render-world] ...)))

2. Title your sections. Title your data definition section, your constant section, and each of your handler sections. Good names for your handler sections are the names of your handlers! Also label your utility functions section.

### Data Definitions

1. Interpret your data. A data definition comes with two parts: the definition and the interpretation. The interpretation should tell the user what your data means in plain English. In some rare cases this interpretation is extraneous. For example, when defining a list of numbers, it would be redundant to write "This represents a list of numbers". Err on the side of caution as it is never wrong to include an interpretation.

2. Remember your examples and templates. Each new data definition should come with its own corresponding examples and template.

### Functions

1. Remember your signature and purpose. Each new function should come with its own corresponding signature and purpose statement. These should be written directly above the function.

2. Write proper signatures. Remember that the only types of data that you can use for a signature are pre-defined types (e.g. String, Number, etc.) or types you have defined yourself. Signatures have one or more input types followed by an arrow and one output type (see above for examples).

3. Write clear purpose statements. A purpose statement should NOT repeat the information given by the signature. It should be succinct and give the reader an understanding of what the function is doing with its inputs. It is good style for the purpose statement of a predicate (a function that outputs a Boolean) to be written as a question.

4. One task, one function. Functions should not be excessively long. If it is, split up the tasks into helper functions.

If a function consumes a complex argument and must perform several different tasks, design several functions that all consume the argument, produce a value in the same data collection, and hand it over to the next function:
 ; HungryHungryHippoWorld -> HungryHungryHippoWorld (define (update-hippo-world hhhw) (update-time (move-marbles (eat-marbles hw)))) ; HungryHungryHippoWorld -> HungryHungryHippoWorld (define (update-time hhhw) ...) ; HungryHungryHippoWorld -> HungryHungryHippoWorld (define (move-marbles hhhw) ...) ; HungryHungryHippoWorld -> HungryHungryHippoWorld (define (eat-marbles hhhw) ...)
Piling the code from these three functions into the first one would yield a confusing mess.

5. Test thoroughly. You should have at LEAST enough tests to cover all your code (none of your code should be highlighted in black when you run it). The exception is any function which calls big-bang as one cannot predict user input. For testing on functions that use randomness, see below.

### General

1. Use names that make sense with respect to the problem, for your data definitions, field names, functions, constants, and parameters.

2. Use proper indentation. Use the indentation style of DrRacket in your program. You can go to "Racket" > "Reindent All" to indent your entire file properly. Press tab to reindent the current line, or the currently selected selected portion of your file.

3. Keep lines narrow. Do not exceed 102 columns for code or comments. DrRacket will show you the current line and column number in the bottom right of the window. You can also use its Edit -> Find Longest Line menu item. Or, you can go to Edit -> Preferences -> Editing -> General Editing, check the Maximum character width guide option and set it to 102.

4. Do not use dangling parentheses: the closing right-parenthesis should be on the same line as the last expression of your code.

 ;; ------------------------ GOOD (define (f l) (cond [(empty? l) 0] [else (f (rest l))])) ;; HERE

 ;; ------------------------ BAD (define (f l) (cond [(empty? 1) 0] [else (f (rest l))] ) ;; NOT HERE )

The dangling parentheses in the second code excerpt are considered extremely bad style.

5. Break lines to break up logically distinct tasks. Consider these examples of simple function calls:
 ;; ----------------- GOOD (define (foo x y z) (max (* x y)      ;; Break after each argument to max, (* y z)      ;; and align all arguments in a column (* x z)      ;; (This works best with short-named functions) (* x y z)))
 ;; ----------------- OK (define (foo x y z) (max          ;; Break after max itself (* x y)      ;; Then indent each argument 1 space (* y z)      ;; (This works better when function names are long) (* x z) (* x y z)))
 ;; ----------------- BAD (define (foo x y z) (max (* x y) (* y z)     ;; This indentation is an inconsistent (* x z)     ;; mix of the previous two styles (* x y z)))
 ;; ----------------- BAD (define (foo x y z) (max (* x y) (* y    ;; This linebreak is just weird. z) (* x z) (* x    ;; This is ugly. And avoidable! y z) ) )

By breaking after each argument, you will more often keep complete expressions together, and need fewer line breaks overall.

In rare cases, you can keep two arguments on a line, when they logically belong together. For example, the x- and y-coordinates in a call to place-image might easily fit on one line, and logically form a pair of coordinates, and so could stay on one line in good style.

Here are some more examples:
 ;; ---------------- BAD (define            ;; Don't break here (foo x y z) ...)
 ;; ---------------- BAD (define-struct      ;; Don't break here foo [x y z]) (define-struct foo  ;; or here [x y z])
 ;; -------------------------- GOOD (define (foo l) (if (some-condition ...) (then-expression ...) (else-expression ...)))
 ;; -------------------------- BAD (define (foo l) (if (some-condition ...) (then-expression ...)    ;; Not aligned with condition (else-expression ...)))
 ;; --------------------------------- VERY GOOD (define (f l)                        ;; Aligning the responses (cond [(empty? l)  0]              ;; in a column is very legible. [else        (f (rest l))])) ;; ...if there's room for it
 ;; --------------------------------- GOOD (define (f l) (cond [(empty? l) 0] [else (f (rest l))]))
 ;; --------------------------------- GOOD (define (f l) (cond [(empty? l) 0] [else (f (rest l))]))
 ;; --------------------------------- OK (define (f l) (cond [(empty? l) 0]                    ;; Only use this style if necessary [else (f (rest l))]))

### Testing & Randomness

When you design functions that create (pseudo-)random results, do not give up on testing. To make this guide concrete, consider this example:

 ; Number Number -> Posn ; create a Posn located at random points in [0,width) x [0,height)
 (define (create-cupcake width height) (make-posn (random width) (random height)))
Just because this function returns a random Posn does not mean you cannot test something about the relationship between its input(s) and output.

At a minimum, your tests can validate that the function always produces a Posn:
 (check-expect (posn? (create-cupcake 100 200)) true) (check-expect (posn? (create-cupcake 20 1000)) true)
Although this simplistic test may appear to be silly, it ensures that the function runs for some examples (e.g., no typos, primitives called correctly).

One step up your tests can include a property checker that confirms the purpose statement:
 ; Number Number Posn -> Boolean ; property tester (check-expect (in-range? 100 200 (create-cupcake 100 200)) true) (check-expect (in-range? 20 1000 (create-cupcake 20 1000)) true) (define (in-range? width height p) (and (<= 0 (posn-x p) width) (<= 0 (posn-y p) height)))
The introduction of a 2-line property checker for a 1-line function may seem overkill, but keep in mind that we purposefully keep the basic example small.

As your functions get more complicated, deploy "loops" to check entire lists, trees, forests and other complex, or randomly generated data structures:Note how one ignores its argument. Still, why would make-list in lieu of build-list not produce the correct result here?

 ; N -> [List-of Posn] ; create the given number of cupcakes (define (create-many-cupcakes n) (local ((define (one _) ; ignore argument, indicate with _ (create-cupcake SCENE-WIDTH SCENE-HEIGHT))) (build-list n one)))

For a function like this, your property checker must use andmap to check all generated Posns:

 ; Posn -> Boolean ; property tester (check-expect (andmap one? (create-many-cupcakes 100)) true) (define (one? p) (in-range? SCENE-WIDTH SCENE-HEIGHT p))