Upgrade to Pro — share decks privately, control downloads, hide ads and more …

最近の研究とか、RubyへのContributionとか / Recent My Study ...

Avatar for TSUYUSATO Kitsune TSUYUSATO Kitsune
August 30, 2025
120

最近の研究とか、RubyへのContributionとか / Recent My Study and Ruby Contributions

RubyKaigi 2025 follow upでの発表スライドです。

https://rhc.connpass.com/event/356128/

Avatar for TSUYUSATO Kitsune

TSUYUSATO Kitsune

August 30, 2025
Tweet

More Decks by TSUYUSATO Kitsune

Transcript

  1. Hiroya Fujinami • g m a kenowjust / t m

    a ke_now_just • Ph.D student a t the N a tion a l Institute of Inform a tics (NII) • Ruby committer, worked on regex m a tching • Ruby 3.2, 3.3: regex m a tching with memoiz a tion 2
  2. RubyKaigi 2025Ͱͷൃද 4 Make Parsers Compatible Using Automata Learning -

    Hiroya Fujinami @ RubyKaigi 2025 Matsuyama Talk Summary 4 I found the compatibility issue (BUG) between Prism and parse.y using automata theory. $ ruby --parser=prism \ 'x = ("a":); p x' :a BUG Applying automata theory, we need to obtain automata corresponding to parsers. For this purpose, we employ automata learning. ΦʔτϚτϯֶशΛ࢖ͬͯɺ ύʔαʔͷόάΛൃݟͨ͠
  3. ATVA 25 6 Intern a tion a l Symposium on

    Autom a ted Technology for Veri f ic a tion a nd An a lysis
  4. Compositional Automata Learning 9 q1 / a component 1 q1

    / a component 2 q1 / a component 3 ෳ਺ͷίϯϙʔωϯτ͕͋ͬͯɺ ίϯϙʔωϯτؒͰ΍ΓऔΓͰ͖Δ γεςϜʹର͢ΔΦʔτϚτϯֶश Moore M a chine Network
  5. Moore Machine Networkͱݱ࣮ͷγεςϜͷؔ܎ 10 q1 / a ΞϓϦέʔγϣϯ αʔόʔ APIαʔόʔ

    1 APIαʔόʔ 2 APIϦΫΤετ Ϩεϙϯε Moore M a chine Network ݱ࣮ੈքͰͷԠ༻ΛΠϝʔδͨ͠ઃఆͷ ݚڀͷͭ΋Γ ϚΠΫϩαʔϏεతͳ
  6. Componentwise Automata Learning 11 q0 / a q1 / b

    p0 / x p1 / x x a b y ෳࡶͳ෦෼ b component 1 component 2 ͜ͷ෦෼ʹ౸ୡ͢Δ͜ͱ͸ ແ͍ ෳ਺ίϯϙʔωϯτΛಉ࣌ʹ ֶश͢Δ͜ͱͰɺ ແବͳ෦෼ΛֶशͤͣʹࡁΉ
  7. SATιϧόΛ࡞Γͨ͘ͳͬͨ • ͷͰ࡞ͬͨ • https://gist.github.com/m a kenowjust/ 254 a 70804179ce3025

    a 274d8c12b5945 • ໊લ͸ KiriMochi • ৭ʑ (CDCL, 2WL, VSIDS, LBD) ࣮૷ͨ͠