12 Years of Ruby

Keynote, RubyConf indonesia 2017

Transcript

1. 12 Years Of Ruby @ponnappa
2. this talk is also about beauty (and magic)
3. but that’s for the end
4. `whoami` formally, Chonira Ponnappa fully, Sidu Ponnappa Chonira Kariappa usually, Sidu
5. LOGO 1991 GWBASIC 1994 QBASIC 1998 C 1999 C++ 1999 Java 2002 C# 2004 Javascript 2005 Ruby 2005
6. started working in 2005
7. 5 startups, 2 jobs, 1 exit
8. at GO-JEK since 2015 managing director, GO-JEK India 2015-2016 head, data engineering 2016-now head, global talent acquisition 2017-now
9. developer founder teacher community organiser sales guy marketing guy hr guy recruiter project manager product manager
10. tl;dr i’ve done a bunch of stuff in 12 years working, and most of it involved Ruby
11. developer founder teacher community organiser sales guy marketing guy hr guy recruiter project manager product manager
12. 2005-2008 activ money and activ mobs ruby 1.8.3/rails 0.11
13. jsp/struts/hibernate vs rails the first aha moment convention > configuration mvc activerecord pattern as a lib! reloading! no more server restarts.
14. unit tests as first class citizens but not TDD, sadly
15. controller tests ftw! functional tests for request/response endpoints filled the gap between unit tests and browser automation
16. automate everything rake + generators
17. ruby was actually slow back then 1.8.x was very slow multi process deployments nginx+lighttpd just restart the process!
18. and i learned rails first (which is a bad idea)
19. minswan easiest community kindest IRC channels ESR’s smart questions FAQ not necessary for survival
20. 2007-now BRUG
21. india’s first major RUG 10 years old this year just under 1000 members
22. my first large community experience 10 years old this year just under 1000 members
23. 2007-2009 rackspace blackbox early private cloud implementation genesis of openstack microservices REST ruby 1.8.7/rails 2.x
24. rails 2.x REST! activeresource (omg, in hindsight it sucked) activerecord query cache json serialization support (phew) plugins (awkward!)
25. phusion passenger + REE finally, a not so painful server setup fork friendly ruby 1.8.x implementation better GC easier on memory
26. importance of specs ruby codebases decay fast migrations from 2.0.x to 2.1.x/2.2.x much pain specs saved the day
27. backward compatibility #fail plugins ecosystem like js lib ecosystem is today
28. microservices rails 2.x made it easy
29. don’t metaprogram cool 2 edged sword more likely to cut me than the problem :) definitely don’t override Object#responds_to? :D
30. rjs server side rendered js another omg #fail ridiculous amounts of coupling so brittle! helped evolve my understand of separation of concerns between frontend and backend
31. did i say REST? oops, but it was (and is) flawed no hypermedia controls a model isn’t a resource actually, a controller represents a resounce level 1 on richardson maturity model
32. still no bundler :( plugins, grrr tried sharing common components across services whattay nightmare
33. 2008-now github.com/kaiwren/wrest fluent REST client lib
34. no good options activeresource net/http libcurl httparty
35. activeresource bolted activerecord semantics over HTTP mapped HTTP verbs to CRUD :/ REST != CRUD JUST DID NOT FIT
36. net/http arrgh that api slow and unreliable no keep alive support totally meh
37. libcurl many rb wrappers different apis but fast and the only sane choice for prod
38. httparty terse api but everything is a class method :/ no client caching
39. so, results of diy multi-runtime support (mri, rbx, jruby, ironruby) (though some are since retired) transparent switch between net/http & libcurl focus on the URI & serialization, not CRUD object oriented & fully specced dev friendly api designed for use with REST endpoints
41. in use across GO-JEK currently handling millions of RPM
42. 2010-now rubyconf india 400+ participants annually
43. first conference learned from pycon india thank you, bang-pypers still going strong genesis of the emerging technology trust
44. led to the ett my first non-profit raises > $100k annually now runs 3 annual open source confereces funds students contributing to oss connects students to oss mentors thank you satish, gautam, niranjan, ajey
45. 2010 c42 engineering boutique ruby/rails consultancy self funded product company $1M run rate in 18 months
46. bundler + rbenv = dependecies multi ruby phew
47. 2010 rubyconf x, new orleans first talk at the main rubyconf
48. open minded look beyond ruby no ego minswan again! such nice people.
49. 2011 github.com/rspec/rspec-mocks any_instance
50. because speccing rails controllers is a pain objects that are create in a controller or lib not everyone does DI mocking non-public apis by reading source is scary long, implementation bound mock chain
51. incredible learning thank you, dchelimsky refactored my PRs, then explained them to me (paitiently!) implementation bound specs anti-pattern
52. 2011-2013 rubymonk.com top ruby e-learning product (still live) ruby 1.9.2/rails 3.x
53. thank you, matz!
54. proved beauty matters the aesthetic made all the difference both visual and narrative style launched at rubyconf xi 8 hours on HN top 3 24 hours on HN homepage
55. empathy taught me to teach through stories objects, blocks, message passing
56. still popular 80-90% of new rubyists use it at some point
57. a labour of love but still, no munny :/ learned the importance of business make it and they will pay is a fallacy
58. 2012-now dev bootcamp taught in ruby
59. consistency of language ruby is more consistent than most popular choices makes it so easy to teach in one of my favourite teaching languages
60. 2015 github.com/opal/opal ruby to js transpiler
61. dream of same language on frontend and backend and that language wasn’t js :)
62. redson github.com/kaiwren/redson my usual obsession with ui and ui frameworks opal based client side component framework for RoR only a poc, because gojek happened
63. in conclusion (and with a healthy dose of hindsight)
64. when i started code was a puzzle that needed to be solved
65. 1991-2005: computer > me my focus was the computer’s productivity
66. BASIC/C/C++/Java making the computer do something was the goal
67. nothing magical about it it was just hard
68. what is magic? write some words perform certain rituals makes the impossible possible
69. that sounds like writing code! it was years before i realised writing code is magic
70. beauty matters in code, beauty is utilitarian beauty makes magic happen
71. utilitarian beauty? easy to understand easy to change easy to scale increasingly magical
72. ruby made me see this ruby was the first language that felt beautiful then i discovered there are others seek them, it’s worth it
73. this was about beauty it was all about making more beautiful things
74. always build beautiful things they just work better much better
75. terima kasih! @ponnappa