## Reference documentation on available primitives

### Data Structures

For memory safety, the following data structures are opaque and only modified
using functions described further down. I still find it useful to understand
how they work under the hood.

- Handles: addresses to objects allocated on the heap. They're augmented with
  book-keeping to guarantee memory-safety, and so cannot be stored in registers.
  See [mu.md](mu.md) for details, but in brief:
    - You need `addr` values to access data they point to.
    - You can't store `addr` values in other types. They're temporary.
    - You can store `handle` values in other types.
    - To convert `handle` to `addr`, use `lookup`.
    - Reclaiming memory (currently unimplemented) invalidates all `addr`
      values.

- Arrays: size-prefixed regions of memory containing multiple elements of a
  single type. Contents are preceded by 4 bytes (32 bits) containing the
  `size` of the array in bytes.

- Slices: a pair of 32-bit addresses denoting a [half-open](https://en.wikipedia.org/wiki/Interval_(mathematics))
  \[`start`, `end`) interval to live memory with a consistent lifetime.

  Invariant: `start` <= `end`

- Streams: strings prefixed by 32-bit `write` and `read` indexes that the next
  write or read goes to, respectively.

  - offset 0: write index
  - offset 4: read index
  - offset 8: size of array (in bytes)
  - offset 12: start of array data

  Invariant: 0 <= `read` <= `write` <= `size`

  By default, writes to a stream abort if it's full. Reads to a stream abort
  if it's empty.

- Graphemes: a sequence of up to 4 utf-8 bytes that encode a single Unicode
  code-point.
- Code-points: integer representing a Unicode character. Must be representable
  in 32 bits as utf-8; largest supported value is 0x10000.

Mu will let you convert between `byte`, `code-point-utf8` and `code-point`
using `copy`, and trust that you know what you're doing. Be aware that doing
so is only correct for English/Latin characters, digits and symbols (ASCII).

### Functions

The most useful functions from 400.mu and later .mu files. Look in
signatures.mu for their full type signatures.

- `abort`: print a message in red on the bottom left of the screen and halt

#### assertions for tests

- `check`: fails current test if given boolean is false (`= 0`).
- `check-not`: fails current test if given boolean isn't false (`!= 0`).
- `check-ints-equal`: fails current test if given ints aren't equal.
- `check-strings-equal`: fails current test if given strings have different bytes.
- `check-stream-equal`: fails current test if stream's data doesn't match
  string in its entirety. Ignores the stream's read index.
- `check-array-equal`: fails if an array's elements don't match what's written
  in a whitespace-separated string.
- `check-next-stream-line-equal`: fails current test if next line of stream
  until newline doesn't match string.

#### predicates

- `handle-equal?`: checks if two handles point at the identical address. Does
  not compare payloads at their respective addresses.

- `array-equal?`: checks if two arrays (of ints only for now) have identical
  elements.

- `string-equal?`: compares two strings.
- `stream-data-equal?`: compares a stream with a string.
- `next-stream-line-equal?`: compares with string the next line in a stream, from
  `read` index to newline.

- `slice-empty?`: checks if the `start` and `end` of a slice are equal.
- `slice-equal?`: compares a slice with a string.
- `slice-starts-with?`: compares the start of a slice with a string.

- `stream-full?`: checks if a write to a stream would abort.
- `stream-empty?`: checks if a read from a stream would abort.

#### arrays

- `populate`: allocates space for `n` objects of the appropriate type.
- `copy-array`: allocates enough space and writes out a copy of an array of
  some type.
- `slice-to-string`: allocates space for an array of bytes and copies the
  slice into it.

#### streams

- `populate-stream`: allocates space in a stream for `n` objects of the
  appropriate type.
- `write-to-stream`: writes arbitrary objects to a stream of the appropriate
  type.
- `read-from-stream`: reads arbitrary objects from a stream of the appropriate
  type.
- `stream-to-array`: allocates just enough space and writes out a stream's
  data between its read index (inclusive) and write index (exclusive).

- `clear-stream`: resets everything in the stream to `0` (except its `size`).
- `rewind-stream`: resets the read index of the stream to `0` without modifying
  its contents.

- `write`: writes a string into a stream of bytes. Doesn't support streams of
  other types.
- `try-write`: writes a string into a stream of bytes if possible. Doesn't
  support streams of other types.
- `write-stream`: concatenates one stream into another.
- `write-slice`: writes a slice into a stream of bytes.
- `append-byte`: writes a single byte into a stream of bytes.
- `append-byte-hex`: writes textual representation of lowest byte in hex to
  a stream of bytes. Does not write a '0x' prefix.
- `read-byte`: reads a single byte from a stream of bytes.
- `read-code-point-utf8`: reads a single unicode code-point-utf8 (up to 4
  bytes) from a stream of bytes.

#### reading/writing hex representations of integers

- `write-int32-hex`
- `hex-int?`: checks if a slice contains an int in hex. Supports '0x' prefix.
- `parse-hex-int`: reads int in hex from string
- `parse-hex-int-from-slice`: reads int in hex from slice
- `parse-array-of-ints`: reads in multiple ints in hex, separated by whitespace.
- `hex-digit?`: checks if byte is in [0, 9] or [a, f] (lowercase only)

- `write-int32-decimal`
- `parse-decimal-int`
- `parse-decimal-int-from-slice`
- `parse-decimal-int-from-stream`
- `parse-array-of-decimal-ints`
- `decimal-digit?`: checks if a code-point-utf8 is in [0, 9]

#### printing to screen

`pixel-on-real-screen` draws a single pixel in one of 256 colors.

All text-mode screen primitives require a screen object, which can be either
the real screen on the computer or a fake screen for tests.

The real screen on the Mu computer can currently display a subset of Unicode.
There is only one font, and it's mostly fixed-width, with individual glyphs
for code-points being either 8 or 16 pixels wide.

- `draw-code-point`: draws a single code-point at a given coordinate, with
  given foreground and background colors. Returns the number of 8-pixel wide
  grid locations used (either 1 or 2).
- `render-code-point`: like `draw-code-point`, but handles newlines and
  updates cursor position. Assumes text is printed left-to-right,
  top-to-bottom.
- `clear-screen`

- `draw-text-rightward`: draws a single line of text, stopping when it reaches
  either the provided bound or the right screen margin.
- `draw-stream-rightward`
- `draw-text-rightward-over-full-screen`: does not provide a bound.
- `draw-text-wrapping-right-then-down`: draws multiple lines of text on screen
  with simplistic word-wrap (no hyphenation) within (x, y) bounds.
- `draw-stream-wrapping-right-then-down`
- `draw-text-wrapping-right-then-down-over-full-screen`
- `draw-int32-hex-wrapping-right-then-down`
- `draw-int32-hex-wrapping-right-then-down-over-full-screen`
- `draw-int32-decimal-wrapping-right-then-down`
- `draw-int32-decimal-wrapping-right-then-down-over-full-screen`

Similar primitives for writing text top-to-bottom, left-to-right.

- `draw-text-downward`
- `draw-stream-downward`
- `draw-text-wrapping-down-then-right`
- `draw-stream-wrapping-down-then-right`
- `draw-text-wrapping-down-then-right-over-full-screen`
- `draw-int32-hex-wrapping-down-then-right`
- `draw-int32-hex-wrapping-down-then-right-over-full-screen`
- `draw-int32-decimal-wrapping-down-then-right`
- `draw-int32-decimal-wrapping-down-then-right-over-full-screen`

Screens remember the current cursor position. The following primitives
automatically read and update the cursor position in various ways.

- `cursor-position`
- `set-cursor-position`
- `draw-code-point-at-cursor-over-full-screen`: `render-code-point` at
  cursor position.
- `draw-cursor`: highlights the current position of the cursor. Programs must
  pass in the code-point to draw at the cursor position, and are responsible
  for clearing the highlight when the cursor moves.
- `move-cursor-left`, `move-cursor-right`, `move-cursor-up`, `move-cursor-down`.
  These primitives always silently fail if the desired movement would go out
  of screen bounds.
- `move-cursor-to-left-margin-of-next-line`
- `move-cursor-rightward-and-downward`: move cursor one code-point-utf8 to the
  right.

- `draw-text-rightward-from-cursor`: truncate at some right margin.
- `draw-text-rightward-from-cursor-over-full-screen`: truncate at right edge
  of screen.
- `draw-text-wrapping-right-then-down-from-cursor`: wrap at some right margin.
- `draw-text-wrapping-right-then-down-from-cursor-over-full-screen`: wrap at
  right edge of screen.
- `draw-int32-hex-wrapping-right-then-down-from-cursor`
- `draw-int32-hex-wrapping-right-then-down-from-cursor-over-full-screen`
- `draw-int32-decimal-wrapping-right-then-down-from-cursor`
- `draw-int32-decimal-wrapping-right-then-down-from-cursor-over-full-screen`

- `draw-text-wrapping-down-then-right-from-cursor`: wrap at some bottom
  margin.
- `draw-text-wrapping-down-then-right-from-cursor-over-full-screen`: wrap at
  bottom edge of screen.

Assertions for tests:

- `check-screen-row`: compare a screen from the left margin of a given row
  index with a string. The row index counts downward from 0 at the top of the
  screen. String can be smaller or larger than a single row, and defines the
  region of interest. Strings longer than a row wrap around to the left margin
  of the next screen row. Currently assumes text is printed left-to-right on
  the screen.
- `check-screen-row-from`: compare a fragment of a screen (left to write, top
  to bottom) starting from a given (x, y) coordinate with an expected string.
  Currently assumes text is printed left-to-right and top-to-bottom on the
  screen.
- `check-screen-row-in-color`: like `check-screen-row` but:
  - also compares foreground color
  - ignores screen locations where the expected string contains spaces
- `check-screen-row-in-color-from`
- `check-screen-row-in-background-color`
- `check-screen-row-in-background-color-from`
- `check-background-color-in-screen-row`: unlike previous functions, this
  doesn't check screen contents, only background color. Ignores background
  color where expected string contains spaces, and compares background color
  where expected string does not contain spaces. Never compares the character
  at any screen location.
- `check-background-color-in-screen-row-from`

#### pixel graphics

- `pixel`: draw a single point at (x, y) with a given color between 0 and 255.
- `draw-line`: between two points (x1, y1) and (x2, y2)
- `draw-horizontal-line`
- `draw-vertical-line`
- `draw-circle`
- `draw-disc`: takes an inner and outer radius
- `draw-monotonic-bezier`: draw curved lines with a single control point.
  Doesn't support curves with "U-turns".

#### events

`read-key` reads a single key from the keyboard and returns it if it exists.
Returns 0 if no key has been pressed. Currently only supports single-byte
(ASCII) keys, which are identical to their code-point and code-point-utf8
representations.

`read-line-from-keyboard` reads keys from keyboard, echoes them to screen
(with given fg/bg colors) and accumulates them in a stream until it encounters
a newline.

`read-mouse-event` returns a recent change in x and y coordinate.

`timer-counter` returns a monotonically increasing counter with some
fixed frequency. You can periodically poll it to check for intervals passing,
but can't make assumptions about how much time has passed.

Mu doesn't currently support interrupt-based events.

We also don't yet have a fake keyboard.

#### persistent storage

`read-ata-disk` synchronously reads a whole number of _sectors_ from a _disk_
of persistent storage. The disk must follow the ATA specification with a
28-bit sector address. Each sector is 512 bytes. Therefore, Mu currently
supports ATA hard disks of up to 128GB capacity.

Similarly, `write-ata-disk` synchronously writes a whole number of sectors to
disk.

Mu doesn't currently support asynchronous transfers to or from a disk.