Slide 1

Slide 1 text

No content

Slide 2

Slide 2 text

@listochkin

Slide 3

Slide 3 text

Ember

Slide 4

Slide 4 text

2018 Looks like Vue Tons of components WASM SSR Preact speeds

Slide 5

Slide 5 text

Awesome?

Slide 6

Slide 6 text

You tell me! No UI in 3? years

Slide 7

Slide 7 text

jQuery React

Slide 8

Slide 8 text

You don’t need a UI

Slide 9

Slide 9 text

Email Calendar CRM Sms Phone API

Slide 10

Slide 10 text

No content

Slide 11

Slide 11 text

Math Physics

Slide 12

Slide 12 text

Programming Languages

Slide 13

Slide 13 text

=> async

Slide 14

Slide 14 text

conditions loops functions data-types

Slide 15

Slide 15 text

Binary Hardware

Slide 16

Slide 16 text

Reality <-> Digital

Slide 17

Slide 17 text

UI

Slide 18

Slide 18 text

Messy Inconsistent Illogical

Slide 19

Slide 19 text

Humans

Slide 20

Slide 20 text

Logical Wise Sentient Decision-makers

Slide 21

Slide 21 text

Emotional Greasy Impulsive Reactionary

Slide 22

Slide 22 text

Math Poetry Music Art

Slide 23

Slide 23 text

Logical and Illogical Consistent and Inconsistent

Slide 24

Slide 24 text

Programmers’ Minefield

Slide 25

Slide 25 text

Human data

Slide 26

Slide 26 text

Text Numbers Time Color & Sound Smell & Taste

Slide 27

Slide 27 text

Logical yet inconsistent

Slide 28

Slide 28 text

2018

Slide 29

Slide 29 text

3 golden rules

Slide 30

Slide 30 text

Store time in UTC

Slide 31

Slide 31 text

Unicode and UTF-8

Slide 32

Slide 32 text

Fixed-point decimals or integers for Money

Slide 33

Slide 33 text

Store time in UTC Unicode and UTF-8 Fixed-point decimals or integers for Money

Slide 34

Slide 34 text

Bad News

Slide 35

Slide 35 text

Bad News #1 Not everybody does it

Slide 36

Slide 36 text

Bad News #2 These rules aren’t enough

Slide 37

Slide 37 text

There was a bank

Slide 38

Slide 38 text

Loans Principal + Interest Monthly Installments

Slide 39

Slide 39 text

SMS Reminder

Slide 40

Slide 40 text

Every month 3 days before payment due at 10:00 am

Slide 41

Slide 41 text

1. Find the right date

Slide 42

Slide 42 text

29th -> 26th

Slide 43

Slide 43 text

Move to 25th for February

Slide 44

Slide 44 text

Except Leap Year

Slide 45

Slide 45 text

Except when that date ends up being a weekend or a bank holiday

Slide 46

Slide 46 text

2. Find the right time

Slide 47

Slide 47 text

10:00 am

Slide 48

Slide 48 text

Time Zones

Slide 49

Slide 49 text

Time Offset vs Time Zone

Slide 50

Slide 50 text

No content

Slide 51

Slide 51 text

No content

Slide 52

Slide 52 text

UTC+2 Finland Ukraine South Africa

Slide 53

Slide 53 text

UTC+3 Finland Ukraine UTC+2 South Africa

Slide 54

Slide 54 text

Time Zones

Slide 55

Slide 55 text

tzdata

Slide 56

Slide 56 text

“Timestamp with timezone” is a lie

Slide 57

Slide 57 text

Timestamp with timezone offset

Slide 58

Slide 58 text

User time = Local DateTime + TimeZone Name

Slide 59

Slide 59 text

Almost nobody does it today

Slide 60

Slide 60 text

Location -> TimeZone

Slide 61

Slide 61 text

APIs Local datasets

Slide 62

Slide 62 text

Moving target

Slide 63

Slide 63 text

tzdata 2018c 10+ releases a year Often bundled with other software

Slide 64

Slide 64 text

App-tzdata Database-tzdata System-tzdata

Slide 65

Slide 65 text

Time Math can be surprising!

Slide 66

Slide 66 text

1. Do it in a single place 2. Update tzdata regularly

Slide 67

Slide 67 text

Numbers

Slide 68

Slide 68 text

Number Base

Slide 69

Slide 69 text

Base 10 0.5 0.111 ⅓

Slide 70

Slide 70 text

Base 10 Financial Data

Slide 71

Slide 71 text

0.1 + 0.2 = 0.30000000000000004

Slide 72

Slide 72 text

True for almost all languages

Slide 73

Slide 73 text

Intel 1977

Slide 74

Slide 74 text

Sign Fraction Exponent

Slide 75

Slide 75 text

(1 + F) * 2^E

Slide 76

Slide 76 text

IEEE 754 - 1985, 2008

Slide 77

Slide 77 text

Money

Slide 78

Slide 78 text

“Use fixed point numbers”

Slide 79

Slide 79 text

+ - *

Slide 80

Slide 80 text

/

Slide 81

Slide 81 text

%

Slide 82

Slide 82 text

Installments

Slide 83

Slide 83 text

(Interest + Premium) / Duration

Slide 84

Slide 84 text

Typical durations: 3, 6, 12, 24, 120, etc

Slide 85

Slide 85 text

Nonoptimal for Base-2

Slide 86

Slide 86 text

Rounding

Slide 87

Slide 87 text

10.265

Slide 88

Slide 88 text

Round-half-up for all installments but the last one

Slide 89

Slide 89 text

10.265 => 10.27

Slide 90

Slide 90 text

Skews the effective rate up Need to compensate

Slide 91

Slide 91 text

Last installment

Slide 92

Slide 92 text

Round-half-to-even Banking rounding

Slide 93

Slide 93 text

10.265 => 10.26 6 is even 7 is add

Slide 94

Slide 94 text

Math.round() Always half-up

Slide 95

Slide 95 text

CSS

Slide 96

Slide 96 text

Text

Slide 97

Slide 97 text

Writing systems Right-to-left Top-to-bottom Graphemes vs Characters

Slide 98

Slide 98 text

6-bit => 7-bit (ASCII) 8-bit CP-1251, koi8-u 16-bit - GB, Big5, Shift-JIS

Slide 99

Slide 99 text

Unicode

Slide 100

Slide 100 text

Let’s encode every writing

Slide 101

Slide 101 text

2 bytes should be enough

Slide 102

Slide 102 text

Windows macOS iOS JavaScript

Slide 103

Slide 103 text

0 - 10FFFFFF

Slide 104

Slide 104 text

UCS-2 => UTF-16 Surrogate pairs

Slide 105

Slide 105 text

1101 10xx xxxx xxxx 1101 11xx xxxx xxxx

Slide 106

Slide 106 text

UTF-8

Slide 107

Slide 107 text

Linux Android Networking Databases

Slide 108

Slide 108 text

iOS device Network Server

Slide 109

Slide 109 text

UTF-8 <=> UTF-16

Slide 110

Slide 110 text

UTF-16 => UTF-8

Slide 111

Slide 111 text

Correct: 2 surrogate pairs Codepoint Re-encode as 3-4 byte UTF-8

Slide 112

Slide 112 text

Practical: Take each surrogate separately 2 codepoints Re-encode as 2 * (3 - 4) byte UTF-8

Slide 113

Slide 113 text

Non-valid UTF-8 bytes CESU-8

Slide 114

Slide 114 text

NULL-byte

Slide 115

Slide 115 text

“Modified UTF-8” Often CESU-8-encoded

Slide 116

Slide 116 text

CESU-8 Oracle MySQL Java Desktop software

Slide 117

Slide 117 text

Forbidden on Web

Slide 118

Slide 118 text

Codepoint vs Character vs Glyph

Slide 119

Slide 119 text

No content

Slide 120

Slide 120 text

U+1F468 U+200D U+1F469 U+200D U+1F467 U+200D U+1F466

Slide 121

Slide 121 text

1 glyph 7 codepoints

Slide 122

Slide 122 text

Cyrillic small letter ha combining breve

Slide 123

Slide 123 text

No content

Slide 124

Slide 124 text

देवनागर Devanagari

Slide 125

Slide 125 text

split substr

Slide 126

Slide 126 text

Unicode-aware regular expressions

Slide 127

Slide 127 text

/\p{Alpha}/u

Slide 128

Slide 128 text

Unicode is not always the answer

Slide 129

Slide 129 text

Han Unification

Slide 130

Slide 130 text

No content

Slide 131

Slide 131 text

Serbian б

Ukrainian б

Slide 132

Slide 132 text

No content

Slide 133

Slide 133 text

Now imagine this for almost every letter

Slide 134

Slide 134 text

Traditional Chinese Simplified Chinese Japanese (Kanji) Korean (Hanja) Vietnamese (Chữ Nôm)

Slide 135

Slide 135 text

No content

Slide 136

Slide 136 text

Plain-text document in 2 CJK languages? Show historic and modern kanji variants? Impossible in Unicode

Slide 137

Slide 137 text

Errors Inaccuracies Ambiguities

Slide 138

Slide 138 text

HTML

Slide 139

Slide 139 text

No Browser API to detect User Input Language

Slide 140

Slide 140 text

window.navigator.languages Statistical methods

Slide 141

Slide 141 text

Sorting

Slide 142

Slide 142 text

Collation

Slide 143

Slide 143 text

Harald Henrik Haakon

Slide 144

Slide 144 text

aa <=> å … x y z æ ø å

Slide 145

Slide 145 text

Sorting is not always character-based

Slide 146

Slide 146 text

kanji 漢字 hiragana (ひらがな / 平仮名) katakana (カタカナ / 片仮名) rōmaji

Slide 147

Slide 147 text

ラドクリフ、マラソン 五輪代表に 1万m出場にも含み

Slide 148

Slide 148 text

Can’t sort text in Kanji

Slide 149

Slide 149 text

No content

Slide 150

Slide 150 text

Search

Slide 151

Slide 151 text

ñ U+006E U+0303 U+00F1

Slide 152

Slide 152 text

Composed Decomposed Canonical Compatibility

Slide 153

Slide 153 text

O ffi ce O f f i ce

Slide 154

Slide 154 text

NKFD Compatible Decomposed

Slide 155

Slide 155 text

Key is to have all strings normalized the same way!

Slide 156

Slide 156 text

ICU

Slide 157

Slide 157 text

International Components for Unicode

Slide 158

Slide 158 text

Iteration Collation Formatting for text and datetimes Normalization

Slide 159

Slide 159 text

~30 MiB

Slide 160

Slide 160 text

JavaScript ECMA-402 Intl

Slide 161

Slide 161 text

Not included in Node by default

Slide 162

Slide 162 text

Be extra careful!

Slide 163

Slide 163 text

No content

Slide 164

Slide 164 text

Every developer made mistakes handling these kinds of data

Slide 165

Slide 165 text

Non-obvious Dormant Painful to find, test, and fix

Slide 166

Slide 166 text

Very expensive bugs

Slide 167

Slide 167 text

Demand knowledge

Slide 168

Slide 168 text

while if Unicode

Slide 169

Slide 169 text

Check your software Teach your team-mates Screen at the interview

Slide 170

Slide 170 text

No content

Slide 171

Slide 171 text

Time Numbers Text @listochkin