Skip to content

Latest commit

 

History

History
221 lines (145 loc) · 4 KB

strings.md

File metadata and controls

221 lines (145 loc) · 4 KB
theme _class paginate backgroundColor backgroundImage style marp
gaia
lead
true
section.photo h1,section.photo h2,section.photo h3,section.photo h4,section.photo h5,section.photo h6 { background-color: #888; color: #FFF; } h6 { font-size: 30%; } img[alt~="centre"] { display: block; margin: 0 auto; }
true

Strings and OsStr: A wild ride through the history of Unicode

Jonathan Pallant

A Journey...

  1. A String is just a String, right?
  2. A Brief History of the String
  3. Not all Strings are alike

A String is just a String, right?

  • String
  • Byte String
  • OS String
  • C Strings

String

let s: String = "Hi 😀!".to_owned();
dbg!(&s);
dbg!(s.len());
dbg!(s.bytes().count());
dbg!(s.chars().count());

▶️

  • A Vector of u8 inside
  • Iterates as 32-bit char

Byte String

let s: [u8; 13] = b"Hello, world!".to_owned();
dbg!(&s);
dbg!(s.len());

▶️

  • Iterates as octets (u8)
  • A Vector of octets (u8) inside

A Brief History of the String

The Punched Card

centre h:500px

Character Encoding

  • Computers work in numbers
  • Humans like to write words
  • Words are made of characters
    • Technically grapheme clusters
    • Is ï one character or two?
  • We need a conversion table!
    • AKA: A Character Set

American Standard Code for Information Interchange

  • Morse Code
  • Telegraph / Baudot codes
  • BCD
  • EBCDIC
  • ASA X3.4-1963
  • aka ASCII

An ASCII Table

centre h:500px

centre h:500px

What if we used the eighth-bit?

  • We get 128 more characters!

centre h:400px

More standards are required...

  • MS-DOS Code Page 437, 850, ...
  • Windows Code Page 1252, 1250, ...
  • Macintosh Code Page 1275, 1282, ...

OK, one Standard to Rule Them All then

Unicode is intended to address the need for a workable, reliable world text encoding. Unicode could be roughly described as "wide-body ASCII" that has been stretched to 16 bits to encompass the characters of all the world's living languages. In a properly engineered design, 16 bits per character are more than sufficient for this purpose.

OK, let's go!

  • Microsoft used it in Windows
  • Sun used it in Java
  • Netscape used it in JavaScript
  • The Standard C Library added wcslen and friends

Unicode 2.0 in 1996...

centre h:400px

  • Unicode Translation Format 16 (UTF-16) arrives

Isn't this the worst of everything?

  • Unit length != number of characters
  • Not ASCII compatible
  • Enter Plan 9 and UTF-8...

UTF-8

  • Variable-length encoding
  • Can encode any Unicode Scalar Value as one, two, three or four bytes.
  • Unit length != number of characters
  • 0b0xxxxxxx
  • 0b110xxxxx 0b10xxxxxx
  • 0b1110xxxx 0b10xxxxxx 0b10xxxxxx

Are we done now?

  • POSIX says file names are an array of 8-bit values
  • Windows says file names are an array of 16-bit wchar_t
  • :(

Not all Strings are alike

  • String/&[str]/"hi"
    • use this by default
  • Vec<u8>/&[u8]/b"hi"
    • use for exchanging data with 8-bit / ASCII systems
  • OsString/OsStr
    • use for exchanging data with your Operating System

C Strings?

  • CString/CStr
    • use for exchanging data with 8-bit C APIs
    • null-terminated
    • Might not be UTF-8
  • https://docs.rs/widestring/
    • use for exchanging data with 'wide' C APIs

Questions?