Pre-evaluation in Ruby - Speaker Deck

Slide 1

Slide 1 text

Pre-evaluation in Ruby

Slide 2

Slide 2 text

Kevin Deisz @kddeisz

Slide 3

Slide 3 text

Warning

Slide 4

Slide 4 text

Compilers

Slide 5

Slide 5 text

Compilers

Slide 6

Slide 6 text

Lexical Analysis Compilers

Slide 7

Slide 7 text

Lexical Analysis Semantic Analysis Compilers

Slide 8

Slide 8 text

Lexical Analysis Semantic Analysis Instruction Generation Compilers

Slide 9

Slide 9 text

Lexical Analysis Semantic Analysis Instruction Generation Optimization Compilers

Slide 10

Slide 10 text

Lexical Analysis Lexical Analytics • Semantic Analysis • Instruction Generation • Optimization

Slide 11

Slide 11 text

Matz is nice so we are nice. Lexical Analytics • Semantic Analysis • Instruction Generation • Optimization

Slide 12

Slide 12 text

Matz is nice so we are nice . Lexical Analytics • Semantic Analysis • Instruction Generation • Optimization

Slide 13

Slide 13 text

is nice so we are nice . Matz Lexical Analytics • Semantic Analysis • Instruction Generation • Optimization

Slide 14

Slide 14 text

is nice so we are nice . Matz Lexical Analytics • Semantic Analysis • Instruction Generation • Optimization

Slide 15

Slide 15 text

Matz is nice we are nice . so Lexical Analytics • Semantic Analysis • Instruction Generation • Optimization

Slide 16

Slide 16 text

Matz is nice so we are nice . Lexical Analytics • Semantic Analysis • Instruction Generation • Optimization

Slide 17

Slide 17 text

Matz is nice so we are nice . Lexical Analytics • Semantic Analysis • Instruction Generation • Optimization

Slide 18

Slide 18 text

Matz is nice so we are . nice Lexical Analytics • Semantic Analysis • Instruction Generation • Optimization

Slide 19

Slide 19 text

Matz is nice so we are nice . Lexical Analytics • Semantic Analysis • Instruction Generation • Optimization

Slide 20

Slide 20 text

Matz is nice so we are nice . Lexical Analytics • Semantic Analysis • Instruction Generation • Optimization

Slide 21

Slide 21 text

Semantic Analysis Lexical Analytics • Semantic Analysis • Instruction Generation • Optimization

Slide 22

Slide 22 text

Matz is nice so we are nice . Lexical Analytics • Semantic Analysis • Instruction Generation • Optimization

Slide 23

Slide 23 text

is nice are nice Lexical Analytics • Semantic Analysis • Instruction Generation • Optimization

Slide 24

Slide 24 text

Grammar Lexical Analytics • Semantic Analysis • Instruction Generation • Optimization

Slide 25

Slide 25 text

verb-phrase : VERB ADJECTIVE Grammar Lexical Analytics • Semantic Analysis • Instruction Generation • Optimization

Slide 26

Slide 26 text

is nice are nice Lexical Analytics • Semantic Analysis • Instruction Generation • Optimization

Slide 27

Slide 27 text

is nice verb phrase verb phrase are nice Lexical Analytics • Semantic Analysis • Instruction Generation • Optimization

Slide 28

Slide 28 text

Matz is nice verb phrase we verb phrase are nice Lexical Analytics • Semantic Analysis • Instruction Generation • Optimization

Slide 29

Slide 29 text

verb-phrase : VERB ADJECTIVE Grammar Lexical Analytics • Semantic Analysis • Instruction Generation • Optimization

Slide 30

Slide 30 text

subject-phrase : NOUN verb-phrase verb-phrase : VERB ADJECTIVE Grammar Lexical Analytics • Semantic Analysis • Instruction Generation • Optimization

Slide 31

Slide 31 text

Matz is nice verb phrase we verb phrase are nice Lexical Analytics • Semantic Analysis • Instruction Generation • Optimization

Slide 32

Slide 32 text

Matz is nice subject phrase subject phrase verb phrase we verb phrase are nice Lexical Analytics • Semantic Analysis • Instruction Generation • Optimization

Slide 33

Slide 33 text

Matz is nice so subject phrase subject phrase verb phrase we verb phrase are nice Lexical Analytics • Semantic Analysis • Instruction Generation • Optimization

Slide 34

Slide 34 text

subject-phrase : NOUN verb-phrase verb-phrase : VERB ADJECTIVE Grammar Lexical Analytics • Semantic Analysis • Instruction Generation • Optimization

Slide 35

Slide 35 text

subordinating-conjunction :  subject-phrase CONJUNCTION subject-phrase subject-phrase : NOUN verb-phrase verb-phrase : VERB ADJECTIVE Grammar Lexical Analytics • Semantic Analysis • Instruction Generation • Optimization

Slide 36

Slide 36 text

Matz is nice so subject phrase subject phrase verb phrase we verb phrase are nice Lexical Analytics • Semantic Analysis • Instruction Generation • Optimization

Slide 37

Slide 37 text

Matz is nice so subord. conjunction subject phrase subject phrase verb phrase we verb phrase are nice Lexical Analytics • Semantic Analysis • Instruction Generation • Optimization

Slide 38

Slide 38 text

Matz is nice so . subord. conjunction subject phrase subject phrase verb phrase we verb phrase are nice Lexical Analytics • Semantic Analysis • Instruction Generation • Optimization

Slide 39

Slide 39 text

Slide 40

Slide 40 text

sentence : subordinating-conjunction PERIOD subordinating-conjunction :  subject-phrase CONJUNCTION subject-phrase subject-phrase : NOUN verb-phrase verb-phrase : VERB ADJECTIVE Grammar Lexical Analytics • Semantic Analysis • Instruction Generation • Optimization

Slide 41

Slide 41 text

Matz is nice so . subord. conjunction subject phrase subject phrase verb phrase we verb phrase are nice Lexical Analytics • Semantic Analysis • Instruction Generation • Optimization

Slide 42

Slide 42 text

Matz is nice so . sentence subord. conjunction subject phrase subject phrase verb phrase we verb phrase are nice Lexical Analytics • Semantic Analysis • Instruction Generation • Optimization

Slide 43

Slide 43 text

Slide 44

Slide 44 text

class Calcp prechigh nonassoc UMINUS left '*' '/' left '+' '-' preclow rule target: exp | /* none */ { result = 0 } exp: exp '+' exp { result += val[2] } | exp '-' exp { result -= val[2] } | exp '*' exp { result *= val[2] } | exp '/' exp { result /= val[2] } | '(' exp ')' { result = val[1] } | '-' NUMBER =UMINUS { result = -val[1] } | NUMBER end https://github.com/tenderlove/racc/blob/master/sample/calc.y Lexical Analytics • Semantic Analysis • Instruction Generation • Optimization

Slide 45

Slide 45 text

| arg '+' arg { $$ = call_bin_op(p, $1, '+', $3, &@2, &@$); } | arg '-' arg { $$ = call_bin_op(p, $1, '-', $3, &@2, &@$); } | arg '*' arg { $$ = call_bin_op(p, $1, '*', $3, &@2, &@$); } | arg '/' arg { $$ = call_bin_op(p, $1, '/', $3, &@2, &@$); } https://github.com/ruby/ruby/blob/trunk/parse.y Lexical Analytics • Semantic Analysis • Instruction Generation • Optimization

Slide 46

Slide 46 text

Slide 47

Slide 47 text

Instruction Generation Lexical Analytics • Semantic Analysis • Instruction Generation • Optimization

Slide 48

Slide 48 text

Matz is nice so . sentence subord. conjunction subject phrase subject phrase verb phrase we verb phrase are nice Lexical Analytics • Semantic Analysis • Instruction Generation • Optimization

Slide 49

Slide 49 text

is nice verb phrase verb phrase are nice Lexical Analytics • Semantic Analysis • Instruction Generation • Optimization

Slide 50

Slide 50 text

verb-phrase : VERB ADJECTIVE  Instructions Lexical Analytics • Semantic Analysis • Instruction Generation • Optimization

Slide 51

Slide 51 text

verb-phrase : VERB ADJECTIVE  push-attr($2) Instructions Lexical Analytics • Semantic Analysis • Instruction Generation • Optimization

Slide 52

Slide 52 text

is nice verb phrase verb phrase are nice Lexical Analytics • Semantic Analysis • Instruction Generation • Optimization

Slide 53

Slide 53 text

Matz is nice subject phrase subject phrase verb phrase we verb phrase are nice Lexical Analytics • Semantic Analysis • Instruction Generation • Optimization

Slide 54

Slide 54 text

verb-phrase : VERB ADJECTIVE  push-attr($2) Grammar Lexical Analytics • Semantic Analysis • Instruction Generation • Optimization

Slide 55

Slide 55 text

subject-phrase : NOUN verb-phrase  verb-phrase : VERB ADJECTIVE  push-attr($2) Grammar Lexical Analytics • Semantic Analysis • Instruction Generation • Optimization

Slide 56

Slide 56 text

subject-phrase : NOUN verb-phrase  save-attr($1, pop-attr) verb-phrase : VERB ADJECTIVE  push-attr($2) Grammar Lexical Analytics • Semantic Analysis • Instruction Generation • Optimization

Slide 57

Slide 57 text

Matz is nice subject phrase subject phrase verb phrase we verb phrase are nice Lexical Analytics • Semantic Analysis • Instruction Generation • Optimization

Slide 58

Slide 58 text

Matz is nice so subord. conjunction subject phrase subject phrase verb phrase we verb phrase are nice Lexical Analytics • Semantic Analysis • Instruction Generation • Optimization

Slide 59

Slide 59 text

subject-phrase : NOUN verb-phrase  save-attr($1, pop-attr) verb-phrase : VERB ADJECTIVE  push-attr($2) Grammar Lexical Analytics • Semantic Analysis • Instruction Generation • Optimization

Slide 60

Slide 60 text

subordinating-conjunction :  subject-phrase CONJUNCTION subject-phrase  subject-phrase : NOUN verb-phrase  save-attr($1, pop-attr) verb-phrase : VERB ADJECTIVE  push-attr($2) Grammar Lexical Analytics • Semantic Analysis • Instruction Generation • Optimization

Slide 61

Slide 61 text

subordinating-conjunction :  subject-phrase CONJUNCTION subject-phrase  cond-skip($1, $2) subject-phrase : NOUN verb-phrase  save-attr($1, pop-attr) verb-phrase : VERB ADJECTIVE  push-attr($2) Grammar Lexical Analytics • Semantic Analysis • Instruction Generation • Optimization

Slide 62

Slide 62 text

Matz is nice so subord. conjunction subject phrase subject phrase verb phrase we verb phrase are nice Lexical Analytics • Semantic Analysis • Instruction Generation • Optimization

Slide 63

Slide 63 text

Matz is nice so . sentence subord. conjunction subject phrase subject phrase verb phrase we verb phrase are nice Lexical Analytics • Semantic Analysis • Instruction Generation • Optimization

Slide 64

Slide 64 text

Slide 65

Slide 65 text

sentence : subordinating-conjunction PERIOD  subordinating-conjunction :  subject-phrase CONJUNCTION subject-phrase  cond-skip($1, $2) subject-phrase : NOUN verb-phrase  save-attr($1, pop-attr) verb-phrase : VERB ADJECTIVE  push-attr($2) Grammar Lexical Analytics • Semantic Analysis • Instruction Generation • Optimization

Slide 66

Slide 66 text

sentence : subordinating-conjunction PERIOD  trace($1) subordinating-conjunction :  subject-phrase CONJUNCTION subject-phrase  cond-skip($1, $2) subject-phrase : NOUN verb-phrase  save-attr($1, pop-attr) verb-phrase : VERB ADJECTIVE  push-attr($2) Grammar Lexical Analytics • Semantic Analysis • Instruction Generation • Optimization

Slide 67

Slide 67 text

Matz is nice so . sentence subord. conjunction subject phrase subject phrase verb phrase we verb phrase are nice Lexical Analytics • Semantic Analysis • Instruction Generation • Optimization

Slide 68

Slide 68 text

Matz is nice so . sentence subord. conjunction subject phrase subject phrase verb phrase we verb phrase are nice trace(5) Lexical Analytics • Semantic Analysis • Instruction Generation • Optimization

Slide 69

Slide 69 text

Matz is nice so subord. conjunction subject phrase subject phrase verb phrase we verb phrase are nice trace(5) Lexical Analytics • Semantic Analysis • Instruction Generation • Optimization

Slide 70

Slide 70 text

Matz is nice so subord. conjunction subject phrase subject phrase verb phrase we verb phrase are nice trace(5) push-attr(“nice”) Lexical Analytics • Semantic Analysis • Instruction Generation • Optimization

Slide 71

Slide 71 text

Matz so subord. conjunction subject phrase subject phrase we verb phrase are nice trace(5) push-attr(“nice”) Lexical Analytics • Semantic Analysis • Instruction Generation • Optimization

Slide 72

Slide 72 text

Matz so subord. conjunction subject phrase subject phrase we verb phrase are nice trace(5) push-attr(“nice”) save-attr(:Matz, pop-attr) Lexical Analytics • Semantic Analysis • Instruction Generation • Optimization

Slide 73

Slide 73 text

so subord. conjunction subject phrase we verb phrase are nice trace(5) push-attr(“nice”) save-attr(:Matz, pop-attr) Lexical Analytics • Semantic Analysis • Instruction Generation • Optimization

Slide 74

Slide 74 text

so subord. conjunction subject phrase we verb phrase are nice trace(5) push-attr(“nice”) save-attr(:Matz, pop-attr) trace(5) push-attr(“nice”) save-attr(:Matz, pop-attr) cond-skip(2) Lexical Analytics • Semantic Analysis • Instruction Generation • Optimization

Slide 75

Slide 75 text

subject phrase we verb phrase are nice trace(5) push-attr(“nice”) save-attr(:Matz, pop-attr) trace(5) push-attr(“nice”) save-attr(:Matz, pop-attr) cond-skip(2) Lexical Analytics • Semantic Analysis • Instruction Generation • Optimization

Slide 76

Slide 76 text

subject phrase we verb phrase are nice trace(5) push-attr(“nice”) save-attr(:Matz, pop-attr) trace(5) push-attr(“nice”) save-attr(:Matz, pop-attr) cond-skip(2) trace(5) push-attr(“nice”) save-attr(:Matz, pop-attr) cond-skip(2) push-attr(“nice”) Lexical Analytics • Semantic Analysis • Instruction Generation • Optimization

Slide 77

Slide 77 text

subject phrase we trace(5) push-attr(“nice”) save-attr(:Matz, pop-attr) trace(5) push-attr(“nice”) save-attr(:Matz, pop-attr) cond-skip(2) trace(5) push-attr(“nice”) save-attr(:Matz, pop-attr) cond-skip(2) push-attr(“nice”) Lexical Analytics • Semantic Analysis • Instruction Generation • Optimization

Slide 78

Slide 78 text

subject phrase we trace(5) push-attr(“nice”) save-attr(:Matz, pop-attr) trace(5) push-attr(“nice”) save-attr(:Matz, pop-attr) cond-skip(2) trace(5) push-attr(“nice”) save-attr(:Matz, pop-attr) cond-skip(2) push-attr(“nice”) trace(5) push-attr(“nice”) save-attr(:Matz, pop-attr) cond-skip(2) push-attr(“nice”) save-attr(:we, pop-attr) Lexical Analytics • Semantic Analysis • Instruction Generation • Optimization

Slide 79

Slide 79 text

trace(5) push-attr(“nice”) save-attr(:Matz, pop-attr) cond-skip(2) push-attr(“nice”) save-attr(:we, pop-attr) Lexical Analytics • Semantic Analysis • Instruction Generation • Optimization

Slide 80

Slide 80 text

Slide 81

Slide 81 text

Optimization Lexical Analytics • Semantic Analysis • Instruction Generation • Optimization

Slide 82

Slide 82 text

Slide 83

Slide 83 text

trace(5) push-attr(“blue”) save-attr(:Sky, pop-attr) cond-skip(2) push-attr(“nice”) save-attr(:we, pop-attr) Lexical Analytics • Semantic Analysis • Instruction Generation • Optimization

Slide 84

Slide 84 text

trace(5) push-attr(“blue”) save-attr(:Sky, pop-attr) cond-skip(2) push-attr(“nice”) save-attr(:we, pop-attr) constant Lexical Analytics • Semantic Analysis • Instruction Generation • Optimization

Slide 85

Slide 85 text

trace(5) push-attr(“nice”) save-attr(:we, pop-attr) Lexical Analytics • Semantic Analysis • Instruction Generation • Optimization

Slide 86

Slide 86 text

trace(5) push-attr(“nice”) save-attr(:we, pop-attr) Lexical Analytics • Semantic Analysis • Instruction Generation • Optimization

Slide 87

Slide 87 text

We are nice. Lexical Analytics • Semantic Analysis • Instruction Generation • Optimization

Slide 88

Slide 88 text

Let’s look at Ruby Lexical Analytics • Semantic Analysis • Instruction Generation • Optimization

Slide 89

Slide 89 text

if 5 + 3 == 8 puts 'Hello, world!' end Lexical Analytics • Semantic Analysis • Instruction Generation • Optimization • Ruby Compilation

Slide 90

Slide 90 text

Slide 91

Slide 91 text

code = -> { if 5 + 3 == 8 puts 'Hello, world!' end } puts RubyVM::InstructionSequence.disasm(code) Lexical Analytics • Semantic Analysis • Instruction Generation • Optimization • Ruby Compilation == disasm: # (catch: FALSE) == catch table | catch type: redo st: 0001 ed: 0023 sp: 0000 cont: 0001 | catch type: next st: 0001 ed: 0023 sp: 0000 cont: 0023 |------------------------------------------------------------------------ 0000 nop ( 1) 0001 putobject 5 ( 2) 0003 putobject 3 0005 opt_plus , ,

Slide 92

Slide 92 text

code = -> { if 5 + 3 == 8 puts 'Hello, world!' end } puts RubyVM::InstructionSequence.disasm(code) == disasm: # (catch: FALSE) == catch table | catch type: redo st: 0001 ed: 0023 sp: 0000 cont: 0001 | catch type: next st: 0001 ed: 0023 sp: 0000 cont: 0023 |------------------------------------------------------------------------ 0000 nop ( 1) 0001 putobject 5 ( 2) 0003 putobject 3 0005 opt_plus , ,

Slide 93

Slide 93 text

Slide 94

Slide 94 text

Slide 95

Slide 95 text

Slide 96

Slide 96 text

Slide 97

Slide 97 text

Slide 98

Slide 98 text

Slide 99

Slide 99 text

if 5 + 3 == 8 puts 'Hello, world!' end Lexical Analytics • Semantic Analysis • Instruction Generation • Optimization • Ruby Compilation

Slide 100

Slide 100 text

if true puts 'Hello, world!' end Lexical Analytics • Semantic Analysis • Instruction Generation • Optimization • Ruby Compilation

Slide 101

Slide 101 text

Lexical Analytics • Semantic Analysis • Instruction Generation • Optimization • Ruby Compilation

Slide 102

Slide 102 text

require 'net/http'; require 'uri'; require 'nokogiri' url = 'https://twitter.com/kddeisz/status/1105570552465575937' body = Net::HTTP.get_response(URI.parse(url)).body node = Nokogiri::HTML(resp).at_css('.tweet-text') eval(node.text) Lexical Analytics • Semantic Analysis • Instruction Generation • Optimization • Ruby Compilation

Slide 103

Slide 103 text

require 'net/http'; require 'uri'; require 'nokogiri' url = 'https://twitter.com/kddeisz/status/1105570552465575937' body = Net::HTTP.get_response(URI.parse(url)).body node = Nokogiri::HTML(resp).at_css('.tweet-text') eval(node.text) 5 + 3 # => 2 Lexical Analytics • Semantic Analysis • Instruction Generation • Optimization • Ruby Compilation

Slide 104

Slide 104 text

Slide 105

Slide 105 text

Slide 106

Slide 106 text

Slide 107

Slide 107 text

WHY WOULD YOU EVER DO THIS!! Lexical Analytics • Semantic Analysis • Instruction Generation • Optimization • Ruby Compilation

Slide 108

Slide 108 text

No content

Slide 109

Slide 109 text

Ripper.sexp_raw('5 + 3') Lexical Analytics • Semantic Analysis • Instruction Generation • Optimization • Ruby Compilation

Slide 110

Slide 110 text

Ripper.sexp_raw('5 + 3') [:program, [:stmts_add, [:stmts_new], [:binary, [:@int, "5", [1, 0]], :+, [:@int, "3", [1, 4]]]]] Lexical Analytics • Semantic Analysis • Instruction Generation • Optimization • Ruby Compilation

Slide 111

Slide 111 text

module Preval class Parser < Ripper::SexpBuilder def self.parse(source) new(source).parse end private SCANNER_EVENTS.each do |event| define_method(:"on_#{event}") do |token| Node.new(:"@#{event}", token, true) end end PARSER_EVENTS.each do |event| define_method(:"on_#{event}") do |*args| Node.new(event, args) end end end end Lexical Analytics • Semantic Analysis • Instruction Generation • Optimization • Ruby Compilation

Slide 112

Slide 112 text

Slide 113

Slide 113 text

Slide 114

Slide 114 text

Slide 115

Slide 115 text

module Preval class Node include Format attr_reader :type, :body, :literal def initialize(type, body, literal = false) @type = type @body = body @literal = literal end def to_source return body if literal public_send(:"to_#{type}") end end end Lexical Analytics • Semantic Analysis • Instruction Generation • Optimization • Ruby Compilation

Slide 116

Slide 116 text

Slide 117

Slide 117 text

module Preval module Format def self.to(type, &block) define_method(:"to_#{type}", &block) end to(:alias) { "alias #{source(0)} #{source(1)}" } to(:aref) { body[1] ? "#{source(0)}[#{source(1)}]" : "#{source(0)}[]" } to(:aref_field) { "#{source(0)}[#{source(1)}]" } to(:arg_paren) { body[0].nil? ? '' : "(#{source(0)})" } to(:args_add) { starts_with?(:args_new) ? source(1) : join(', ') } to(:args_add_block) do args, block = body parts = args.is?(:args_new) ? [] : [args.to_source] parts << "#{parts.any? ? ', ' : ''}{block.to_source}" if block parts.join end to(:args_add_star) { starts_with?(:args_new) ? "*#{source(1)}" : "#{source(0)}, *#{source(1)}" } to(:args_new) { '' } to(:assign) { "#{source(0)} = #{source(1)}" } to(:array) { body[0].nil? ? '[]' : "#{starts_with?(:args_add) ? '[' : ''}#{source(0)}]" } to(:assoc_new) { starts_with?(:@label) ? join(' ') : join(' => ') } to(:assoc_splat) { "**#{source(0)}" } to(:assoclist_from_args) { body[0].map(&:to_source).join(', ') } to(:bare_assoc_hash) { body[0].map(&:to_source).join(', ') } to(:begin) { "begin\n#{join("\n")}\nend" } to(:BEGIN) { "BEGIN {\n#{source(0)}\n}"} to(:binary) { "#{source(0)} #{body[1]} #{source(2)}" } to(:blockarg) { "{source(0)}" } to(:block_var) { "|#{source(0)}|" } to(:bodystmt) do source(0).tap do |code| code << "\n#{source(1)}" if body[1] if body[2] stmts = body[2].is?(:else) ? body[2].body[0] : body[2] code << "\nelse\n#{stmts.to_source}" end Lexical Analytics • Semantic Analysis • Instruction Generation • Optimization • Ruby Compilation

Slide 118

Slide 118 text

to(:string_literal) do content = source(0).map! do |part| part.start_with?('#{') ? part : part.gsub(/([^\\])"/) { "#{$1}\\\"" } end "\"#{content.join}\"" end to(:super) { "super#{starts_with?(:arg_paren) ? '' : ' '}#{source(0)}" } to(:symbol) { ":#{source(0)}" } to(:symbol_literal) { source(0) } to(:symbols_add) { join(starts_with?(:symbols_new) ? '' : ' ') } to(:symbols_new) { '%I[' } to(:top_const_field) { "::#{source(0)}" } to(:top_const_ref) { "::#{source(0)}" } to(:unary) { "#{body[0][0]}#{source(1)}" } to(:undef) { "undef #{body[0][0].to_source}" } to(:unless) { "unless #{source(0)}\n#{source(1)}\n#{body[2] ? "#{source(2)}\n" : ''}end" } to(:unless_mod) { "#{source(1)} unless #{source(0)}" } to(:until) { "until #{source(0)}\n#{source(1)}\nend" } to(:until_mod) { "#{source(1)} until #{source(0)}" } to(:var_alias) { "alias #{source(0)} #{source(1)}" } to(:var_field) { join } to(:var_ref) { source(0) } to(:vcall) { join } to(:void_stmt) { '' } to(:when) { "when #{source(0)}\n#{source(1)}#{body[2] ? "\n#{source(2)}" : ''}" } to(:while) { "while #{source(0)}\n#{source(1)}\nend" } to(:while_mod) { "#{source(1)} while #{source(0)}" } to(:word_add) { join } to(:word_new) { '' } to(:words_add) { join(starts_with?(:words_new) ? '' : ' ') } to(:words_new) { '%W[' } to(:xstring_add) { join } to(:xstring_new) { '' } to(:xstring_literal) { "%x[#{source(0)}]" } to(:yield) { "yield#{starts_with?(:paren) ? '' : ' '}#{join}" } to(:yield0) { 'yield' } to(:zsuper) { 'super' } end end Lexical Analytics • Semantic Analysis • Instruction Generation • Optimization • Ruby Compilation

Slide 119

Slide 119 text

module Preval class << self attr_reader :visitors def enable_all! Visitors::Arithmetic.enable! Visitors::AttrAccessor.enable! Visitors::Fasterer.enable! Visitors::Loops.enable! end def process(source) visitors.inject(Parser.parse(source)) do |current, visitor| current.tap { |ast| ast.visit(visitor) } end.to_source end end @visitors = [] end Lexical Analytics • Semantic Analysis • Instruction Generation • Optimization • Ruby Compilation

Slide 120

Slide 120 text

Slide 121

Slide 121 text

Slide 122

Slide 122 text

module Preval class Node def visit(pass) return if literal handler = :"on_#{type}" pass.public_send(handler, self) if pass.respond_to?(handler) return unless body.is_a?(Array) body.each do |child| child.visit(pass) if child.is_a?(Node) end end end end Lexical Analytics • Semantic Analysis • Instruction Generation • Optimization • Ruby Compilation

Slide 123

Slide 123 text

class Preval::Visitors::Arithmetic < Visitor OPERATORS = %i[+ - * / % **].freeze def on_binary(node) left, operation, right = node.body return unless OPERATORS.include?(operation) if left.is?(:@int) && right.is?(:@int) value = left.to_int.public_send(operation, right.to_int) node.update(:@int, value.to_s) end end end Lexical Analytics • Semantic Analysis • Instruction Generation • Optimization • Ruby Compilation

Slide 124

Slide 124 text

Demo Lexical Analytics • Semantic Analysis • Instruction Generation • Optimization • Ruby Compilation

Slide 125

Slide 125 text

if defined?(Bootsnap) load_iseq = RubyVM::InstructionSequence.method(:load_iseq) if load_iseq.source_location[0].include?('/bootsnap/') Bootsnap::CompileCache::ISeq.singleton_class.prepend( Module.new do def input_to_storage(source, path) source = Preval.process(source) compiled = RubyVM::InstructionSequence.compile( source, path, path ) compiled.to_binary rescue SyntaxError raise Bootsnap::CompileCache::Uncompilable end end ) end end Lexical Analytics • Semantic Analysis • Instruction Generation • Optimization • Ruby Compilation

Slide 126

Slide 126 text

Slide 127

Slide 127 text

Slide 128

Slide 128 text

Slide 129

Slide 129 text

Slide 130

Slide 130 text

Slide 131

Slide 131 text

Slide 132

Slide 132 text

Slide 133

Slide 133 text

Slide 134

Slide 134 text

Lexical Analytics • Semantic Analysis • Instruction Generation • Optimization • Ruby Compilation

Slide 135

Slide 135 text

Optimizations Lexical Analytics • Semantic Analysis • Instruction Generation • Optimization • Ruby Compilation

Slide 136

Slide 136 text

that can be done now Optimizations Lexical Analytics • Semantic Analysis • Instruction Generation • Optimization • Ruby Compilation

Slide 137

Slide 137 text

that can be done now  attr_accessor replacement Optimizations Lexical Analytics • Semantic Analysis • Instruction Generation • Optimization • Ruby Compilation

Slide 138

Slide 138 text

that can be done now  attr_accessor replacement  instruction elimination Optimizations Lexical Analytics • Semantic Analysis • Instruction Generation • Optimization • Ruby Compilation

Slide 139

Slide 139 text

that can be done now  attr_accessor replacement  instruction elimination that can be done with deoptimization Optimizations Lexical Analytics • Semantic Analysis • Instruction Generation • Optimization • Ruby Compilation

Slide 140

Slide 140 text

that can be done now  attr_accessor replacement  instruction elimination that can be done with deoptimization  for-loop replacement Optimizations Lexical Analytics • Semantic Analysis • Instruction Generation • Optimization • Ruby Compilation

Slide 141

Slide 141 text

that can be done now  attr_accessor replacement  instruction elimination that can be done with deoptimization  for-loop replacement  constant folding Optimizations Lexical Analytics • Semantic Analysis • Instruction Generation • Optimization • Ruby Compilation

Slide 142

Slide 142 text

that can be done now  attr_accessor replacement  instruction elimination that can be done with deoptimization  for-loop replacement  constant folding that a compiler can never do Optimizations Lexical Analytics • Semantic Analysis • Instruction Generation • Optimization • Ruby Compilation

Slide 143

Slide 143 text

that can be done now  attr_accessor replacement  instruction elimination that can be done with deoptimization  for-loop replacement  constant folding that a compiler can never do  replacing certain APIs with newer or faster APIs Optimizations Lexical Analytics • Semantic Analysis • Instruction Generation • Optimization • Ruby Compilation

Slide 144

Slide 144 text

But why? Lexical Analytics • Semantic Analysis • Instruction Generation • Optimization • Ruby Compilation

Slide 145

Slide 145 text

Style Progression Lexical Analytics • Semantic Analysis • Instruction Generation • Optimization • Ruby Compilation

Slide 146

Slide 146 text

Nothing Style Progression Lexical Analytics • Semantic Analysis • Instruction Generation • Optimization • Ruby Compilation

Slide 147

Slide 147 text

Nothing Senior devs enforce in code review Style Progression Lexical Analytics • Semantic Analysis • Instruction Generation • Optimization • Ruby Compilation

Slide 148

Slide 148 text

Nothing Senior devs enforce in code review Style guide is developed and linked Style Progression Lexical Analytics • Semantic Analysis • Instruction Generation • Optimization • Ruby Compilation

Slide 149

Slide 149 text

Nothing Senior devs enforce in code review Style guide is developed and linked Linters are developed and run locally (and in CI) Style Progression Lexical Analytics • Semantic Analysis • Instruction Generation • Optimization • Ruby Compilation

Slide 150

Slide 150 text

Nothing Senior devs enforce in code review Style guide is developed and linked Linters are developed and run locally (and in CI) Auto-formatters and compilers take care of it Style Progression Lexical Analytics • Semantic Analysis • Instruction Generation • Optimization • Ruby Compilation

Slide 151

Slide 151 text

Nothing Senior devs enforce in code review Style guide is developed and linked Linters are developed and run locally (and in CI) Auto-formatters and compilers take care of it Style Progression w e’re here Lexical Analytics • Semantic Analysis • Instruction Generation • Optimization • Ruby Compilation

Slide 152

Slide 152 text

language: ruby rvm: 2.6.3 branches: only: master before_install: - gem update --system - gem install bundler before_script: bundle exec rails db:create cache: bundler script: - bundle exec bundle audit - bin/rails test - bundle exec rubocop --parallel - bin/rails reek - bundle exec brakeman -z - bin/rails fasterer after_success: bin/deploy services: - postgresql - redis Lexical Analytics • Semantic Analysis • Instruction Generation • Optimization • Ruby Compilation

Slide 153

Slide 153 text

Slide 154

Slide 154 text

Slide 155

Slide 155 text

Slide 156

Slide 156 text

Slide 157

Slide 157 text

Slide 158

Slide 158 text

Slide 159

Slide 159 text

language: ruby rvm: 2.6.3 branches: only: master before_install: - gem update --system - gem install bundler before_script: bundle exec rails db:create cache: bundler script: - bin/rails test after_success: bin/deploy services: - postgresql - redis Lexical Analytics • Semantic Analysis • Instruction Generation • Optimization • Ruby Compilation

Slide 160

Slide 160 text

Slide 161

Slide 161 text

Auto-formatters and compilers take care of it Style Progression Lexical Analytics • Semantic Analysis • Instruction Generation • Optimization • Ruby Compilation

Slide 162

Slide 162 text

Lexical Analytics • Semantic Analysis • Instruction Generation • Optimization • Ruby Compilation

Slide 163

Slide 163 text

Write code Lexical Analytics • Semantic Analysis • Instruction Generation • Optimization • Ruby Compilation

Slide 164

Slide 164 text

Write code Test code Lexical Analytics • Semantic Analysis • Instruction Generation • Optimization • Ruby Compilation

Slide 165

Slide 165 text

Write code Test code Deploy code Lexical Analytics • Semantic Analysis • Instruction Generation • Optimization • Ruby Compilation

Slide 166

Slide 166 text

Pre-evaluation in Ruby github.com/kddeisz/preval Kevin Deisz @kddeisz