Ruby Internals V3

How Ruby code gets executed by MRI

Transcript

1. michelada.io Ruby Internals Mario Alberto Chávez @mario_chavez
2. michelada.io michelada.io México
3. michelada.io michelada.io México michelada.io Colima
4. michelada.io michelada.io michelada.io
5. michelada.io michelada.io michelada.io
6. michelada.io Exploraremos la version C de Ruby
7. michelada.io Teoría de la computación
8. michelada.io Qué pasa cuando Ruby ejecuta nuestro código? irb(main): 001:0> 2 + 2 => 4
9. michelada.io Tokenizer
10. michelada.io Categorizar una cadena de texto en elementos que Ruby comprenda
11. michelada.io 10.times do |i| puts i end
12. michelada.io on_int 10 . on_ident “times” on_kw “do” | on_ident “i” | on_ident “puts” on_ident “i” on_kw “end”
13. michelada.io Cómo sucede la tokenización?
14. michelada.io Autómatas Finitos q0 q1 q2 a a a b b b Start Final state
15. michelada.io Lex es un generador de análisis léxico
16. michelada.io Ruby tiene su propio Lexer
17. michelada.io parse.y contiene el lexer de Ruby parser_yylex(struct parser_params *parser)
18. michelada.io Análisis léxico con Ripper
19. michelada.io require ‘ripper’ require ‘pp’ code = <<STR 
 10.times do |i| puts i end
 STR
 
 pp Ripper.lex(code)
20. michelada.io [[[1, 0], :on_int, "10"], [[1, 2], :on_period, "."], [[1, 3], :on_ident, "times"], [[1, 8], :on_sp, " "], [[1, 9], :on_kw, "do"], [[1, 11], :on_sp, " "], [[1, 12], :on_op, "|"], [[1, 13], :on_ident, "i"], [[1, 14], :on_op, "|"], [[1, 15], :on_ignored_nl, "\n"], [[2, 0], :on_ident, "puts"], [[2, 4], :on_sp, " "], [[2, 5], :on_ident, "i"], [[2, 6], :on_nl, "\n"], [[3, 0], :on_kw, "end"], [[3, 3], :on_nl, "\n"]]
21. michelada.io Parser
22. michelada.io Convertir los tokens a grupos de sentencias que Ruby entienda
23. michelada.io 10.times do|i| puts i end var ref command args add 
 block indetifier “puts” indetifier “i”
24. michelada.io Cómo sucede el parsing?
25. michelada.io Ruby usa Bison
26. michelada.io Bison genera parsers LALR
27. michelada.io parse.y contiene las reglas gramaticales de Ruby
28. michelada.io expr : command_call | expr keyword_and expr { $$ = logop(idAND, $1, $3); } | expr keyword_or expr { $$ = logop(idOR, $1, $3); }
29. michelada.io Análisis gramatical con Ripper
30. michelada.io require ‘ripper’ require ‘pp’ code = <<STR 10.times do |i| puts i end STR pp Ripper.sexp(code)
31. michelada.io [:program, [[:method_add_block, [:call, [:@int, "10", [1, 0]], :".", [:@ident, "times", [1, 3]]], [:do_block, [:block_var, [:params, [[:@ident, "i", [1, 13]]], nil, nil, nil, nil, nil, nil], false], [[:command, [:@ident, "puts", [2, 2]], [:args_add_block, [[:var_ref, [:@ident, "i", [2, 7]]]], false]]]]]]]
32. michelada.io A partir del parse, es posible crear un AST
33. michelada.io # my_program.rb
 puts 2 + 2
 
 ruby --dump parsetree my_program.rb
34. michelada.io # | @ NODE_FCALL (line: 1) # | +- nd_mid: :puts # | +- nd_args: # | @ NODE_ARRAY (line: 1) # | +- nd_alen: 1 # | +- nd_head: # | | @ NODE_CALL (line: 1) # | | +- nd_mid: :+ # | | +- nd_recv: # | | | @ NODE_LIT (line: 1) # | | | +- nd_lit: 2 # | | +- nd_args: # | | @ NODE_ARRAY (line: 1) # | | +- nd_alen: 1 # | | +- nd_head: # | | | @ NODE_LIT (line: 1) # | | | +- nd_lit: 2 # | | +- nd_next: # | | (null node)
35. michelada.io puts 2 + 2 NODE_FCALL method_id: “puts” NODE_CALL method_id: “+” NODE_LITERAL “2” NODE_LITERAL “2”
36. michelada.io Compile
37. michelada.io Convertir el AST en instrucciones para el YARV
38. michelada.io Ruby 1.9 introduce YARV
39. michelada.io Ruby cuenta con las herramientas para compilar al YARV
40. michelada.io code = <<STR Puts 2 + 2 STR pp RubyVM::InstructionSequence.compile( code).disasm
41. michelada.io 0000 trace 1 ( 1) 0002 putself 0003 putobject 2 0005 putobject 2 0007 opt_plus <callinfo!mid:+, argc:1, ARGS_SIMPLE>, <callcache> 0010 opt_send_without_block <callinfo!mid:puts, argc:1, FCALL|ARGS_SIMPLE>, <callcache> 0013 leave
42. michelada.io puts 2 + 2 NODE_FCALL method_id: “puts” NODE_CALL method_id: “+” NODE_LITERAL “2” NODE_LITERAL “2” INSTRUCCIONES YARV putself putobject 2 putobject 2 send :+, 1 send :puts, 1 opt_plus
43. michelada.io El archivo insns.def contiene las instrucciones de YARV
44. michelada.io Cómo se ejecuta el código?
45. michelada.io YARV es una VM de stack
46. michelada.io puts 2 + 2 rb_control_frame_t Instrucciones YARV
 putself putobject 2 putobject 2 opt_plus send :puts, 1 stack interno de YARV 
 self rb_control_frame_t PC SP self type SP PC rb_control_frame_t 2 2 4
47. michelada.io insns.def contiene las instrucciones del YARV
48. michelada.io opt_plus (CALL_INFO ci, CALL_CACHE cc) (VALUE recv, VALUE obj) (VALUE val) { val = vm_opt_plus(recv, obj); if (val == Qundef) { /* other */ PUSH(recv); PUSH(obj); CALL_SIMPLE_METHOD(recv); } }
49. michelada.io Finalmente nuestro código es ejecutado
50. michelada.io Gracias! Mario Alberto Chávez @mario_chavez