Video and slides synchronized, mp3 and slide download available at URL http://bit.ly/2Cs3wyC. Carmen Andoh provides the historical context around the technical decisions of the Go language to better understand its concurrency primitives, garbage collection, and small standard library. Filmed at qconsf.com. Carmen Andoh is a software engineer on the Build Infrastructure team at Travis CI. She was the first scholarship recipient for Gophercon in 2015, where she was first introduced to Go, and hasn't looked back.

1. The Why of Go21st Century Programming Languages Track, Qcon SF
Carmen Andoh

2. InfoQ.com: News & Community Site
• Over 1,000,000 software developers, architects and CTOs read the site world-
wide every month
• 250,000 senior developers subscribe to our weekly newsletter
• Published in 4 languages (English, Chinese, Japanese and Brazilian
Portuguese)
• Post content from our QCon conferences
• 2 dedicated podcast channels: The InfoQ Podcast, with a focus on
Architecture and The Engineering Culture Podcast, with a focus on building
• 96 deep dives on innovative topics packed as downloadable emags and
minibooks
• Over 40 new content items per week
Watch the video with slide
synchronization on InfoQ.com!
https://www.infoq.com/presentations/
go-concurrency-gc

3. Purpose of QCon
- to empower software development by facilitating the spread of
knowledge and innovation
Strategy
- practitioner-driven conference designed for YOU: influencers of
change and innovation in your teams
- speakers and topics driving the evolution and innovation
- connecting and catalyzing the influencers and innovators
Highlights
- attended by more than 12,000 delegates since 2007
- held in 9 cities worldwide
Presented at QCon San Francisco
www.qconsf.com

4. 1983

5. ● Dave Cheney
● Alan Donovan
● Steve Francia
● Jérôme Pettazoni
Thank you, fellow time travelers from the Future

6. Back to 1983 … (from 1985, but whetev, 80s rule)

11. Genealogy of Programming Languages
1956-1983
You
are
here

12. Genealogy of
Programming
Languages
1956-2015

13. You
are
here

15. Genealogy of
Programming
Languages
1956-2015

20. GNUGNU’s Not Unix Project
Started in 1983

29. ● too simple / lack of syntactic sugar
● no generics
● bad dependency management
● stuck in 70/80's
● error handling
● no unused imports
● too opinionated
● too verbose
● no ternary operator
● no macros or
templates
https://github.com/ksimka/go-is-not-good

32. “Go programming language was conceived as an answer to
some of the problems we were seeing developing software
infrastructure at Google. The computing landscape today is
almost unrelated to the environment in which the languages
being used, mostly C++, Java, and Python, had been created.
The problems introduced by multicore processors,
networked systems, massive computation clusters, and
the web programming model were being worked around
rather than addressed head-on.

33. “Go programming language was conceived as an answer to
some of the problems we were seeing developing software
infrastructure at Google. The computing landscape today is
almost unrelated to the environment in which the languages
being used, mostly C++, Java, and Python, had been created.
The problems introduced by multicore processors,
networked systems, massive computation clusters, and
the web programming model were being worked around
rather than addressed head-on.

34. “Go programming language was conceived as an answer to
some of the problems we were seeing developing software
infrastructure at Google. The computing landscape today is
almost unrelated to the environment in which the languages
being used, mostly C++, Java, and Python, had been created.
The problems introduced by multicore processors,
networked systems, massive computation clusters, and
the web programming model were being worked around
rather than addressed head-on.

35. “Go programming language was conceived as an answer to
some of the problems we were seeing developing software
infrastructure at Google. The computing landscape today is
almost unrelated to the environment in which the languages
being used, mostly C++, Java, and Python, had been created.
The problems introduced by multicore processors,
networked systems, massive computation clusters, and
the web programming model were being worked around
rather than addressed head-on.

36. “Go programming language was conceived as an answer to
some of the problems we were seeing developing software
infrastructure at Google. The computing landscape today is
almost unrelated to the environment in which the languages
being used, mostly C++, Java, and Python, had been created.
The problems introduced by multicore processors,
networked systems, massive computation clusters, and
the web programming model were being worked around
rather than addressed head-on.

37. Moreover, the scale has changed: today’s server programs
comprise tens of millions of lines of code, are worked on by
hundreds or even thousands of programmers, and are
updated literally every day. To make matters worse, build time
have stretched to many minutes, even hours, even
languages,on large compilation clusters,”

38. Moreover, the scale has changed: today’s server programs
comprise tens of millions of lines of code, are worked on by
hundreds or even thousands of programmers, and are
updated literally every day. To make matters worse, build time
has stretched to many minutes, even hours, on large
compilation clusters”

39. Moreover, the scale has changed: today’s server programs
comprise tens of millions of lines of code, are worked on by
hundreds or even thousands of programmers, and are
updated literally every day. To make matters worse, build time
has stretched to many minutes, even hours, on large
compilation clusters”

40. ● multicore processors
● networked systems
● massive computation
clusters
● web programming model
● hundreds or even
thousands of programmers
● large compilation clusters

41. 20172000 2005 201090s80s
Go

42. 20172000 2005 201090s80s
Software|
Languages
Go

43. 20172000 2005 201090s80s
Software|
Languages
Hardware&
Compute
Go

44. 20172000 2005 201090s80s
Software|
Languages
Hardware&
Compute
Context
Go

46. Hardware&
Compute
Software|
Languages
Context
20172000 2005 201090s80s
Go
UTF-8
Unix
(60s)
C (68-73)
Robert Griesemer, Rob Pike, Ken Thompson

48. Hardware&
Compute
Software|
Languages
Context
20172000 2005 201090s80s
Go
UTF-8
Unix
(60s)
C (68-73)
Plan 9
OS
Hotspot
JVM
Robert Griesemer, Rob Pike, Ken Thompson

49. ● Concurrency
● Distributed Systems
● Garbage Collection
● Memory Locality
● Readability
Go’s 21st Century Characteristics

50. 20172000 2005 201090s80s
Software|
Languages
Hardware&
Compute
Context
Go
Concurrency
C++ at
Google

51. 20172000 2005 201090s80s
Software|
Languages
Hardware&
Compute
Context
Go
Concurrency
C++ at
Google
C10K

52. 20172000 2005 201090s80s
Software|
Languages
Hardware&
Compute
Context
Go
Concurrency
C++ at
Google
C10K
Epoll
(linux)
Kqueue
(BSD)

53. 20172000 2005 201090s80s
Software|
Languages
Hardware&
Compute
Context
Go
Concurrency
C++ at
Google
C10K
Epoll
(linux)
Kqueue
(BSD)
SEDA
Java NIO
P4
Core 2

54. 20172000 2005 201090s80s
Software|
Languages
Hardware&
Compute
Context
Go
Concurrency
C++ at
Google
C10K
Epoll
(linux)
Kqueue
(BSD)
SEDA
Java NIO

55. 20172000 2005 201090s80s
Software|
Languages
Hardware&
Compute
Context
Go
Concurrency
C++ at
Google
C10K
Epoll
(linux)
Kqueue
(BSD)
SEDA
Java NIO
P4
Core 2

56. Hardware&
Compute
Software|
Languages
Context
20172000 2005 201090s80s
Go

57. 20172000 2005 201090s80s
Software|
Languages
Hardware&
Compute
Context
Go
Concurrency
C++ at
Google
C10K
Epoll
(linux)
Kqueue
(BSD)
SEDA
Java NIO
P4
Core 2

58. 20172000 2005 201090s80s
Software|
Languages
Hardware&
Compute
Context
Go
Concurrency
C++ at
Google
C10K
Epoll
(linux)
Kqueue
(BSD)
SEDA
Java NIO
P4
Core 2
Ruby
Mongrel
Gunicorn
(Python)

59. Hardware&
Compute
Software|
Languages
Context
20172000 2005 201090s80s
Go

60. Hardware&
Compute
Software|
Languages
Context
20172000 2005 201090s80s
Go

61. Events, Threads and Goroutines
Nginx - event loop plus state machine model

62. Events, Threads and Goroutines
Nginx - event loop plus state machine model
App logic

63. Events, Threads and Goroutines
Nginx - event loop plus state machine model
App logic

64. Events, Threads and Goroutines
Nginx - event loop plus state machine model
App

65. Events, Threads and Goroutines
Nginx - event loop plus state machine model
App

66. Events, Threads and Goroutines
Nginx - event loop plus state machine model
App

67. Events, Threads and Goroutines
Nginx - event loop plus state machine model
App

68. Events, Threads and Goroutines
Nginx - event loop plus state machine model
App

69. Events, Threads and Goroutines
Nginx - event loop plus state machine model
App

70. 20172000 2005 201090s80s
Software|
Languages
Hardware&
Compute
Context
Go
Concurrency
C++ at
Google
C10K
Epoll
(linux)
Kqueue
(BSD)
SEDA
Java NIO
P4
Core 2
Ruby
Mongrel
Gunicorn
(Python)
NodeJS

71. 20172000 2005 201090s80s
Software|
Languages
Hardware&
Compute
Context
Go
Concurrency
C++ at
Google
C10K
Epoll
(linux)
Kqueue
(BSD)
SEDA
Java NIO
P4
Core 2
Ruby
Mongrel
Gunicorn
(Python)
NodeJS

72. 20172000 2005 201090s80s
Software|
Languages
Hardware&
Compute
Context
Go
Concurrency
C++ at
Google
C10K
Epoll
(linux)
Kqueue
(BSD)
SEDA
Java NIO
P4
Core 2
Ruby
Mongrel
Gunicorn
(Python)
NodeJS

78. App

79. App
= Threads

83. 20172000 2005 201090s80s
Software|
Languages
Hardware&
Compute
Context
Go
Distributed
Systems
C++ at
Google
C10K
Epoll
(linux)
Kqueue
(BSD)
SEDA
Java NIO
P4
Core 2
Ruby
Mongrel
Gunicorn
(Python)
NodeJS
The rise of
containers +
cloud-native+
ecosystem
Microservices
serverless

84. 20172000 2005 201090s80s
Software|
Languages
Hardware&
Compute
Context
Go
Garbage
Collection
C++ at
Google
C10K
Epoll
(linux)
Kqueue
(BSD)
SEDA
Java NIO
P4
Core 2
Ruby
Mongrel
Gunicorn
(Python)
NodeJS
The rise of
containers +
cloud-native+
ecosystem
Microservices
serverless

85. 20172000 2005 201090s80s
Software|
Languages
Hardware&
Compute
Context
Go
Memory
Locality
C++ at
Google
C10K
Epoll
(linux)
Kqueue
(BSD)
SEDA
Java NIO
P4
Core 2
Ruby
Mongrel
Gunicorn
(Python)
NodeJS
The rise of
containers +
cloud-native+
ecosystem
Microservices
serverless

86. 20172000 2005 201090s80s
Software|
Languages
Hardware&
Compute
Context
Go
Memory
Locality
C++ at
Google
C10K
Epoll
(linux)
Kqueue
(BSD)
SEDA
Java NIO
P4
Core 2
Ruby
Mongrel
Gunicorn
(Python)
NodeJS
The rise of
containers +
cloud-native+
ecosystem
Microservices
serverless
Hotspot
JVM

87. Memory Locality

88. Memory Locality
Java

89. Memory Locality
Java
No value types
Everything Allocated

90. Memory Locality
Java
No value types
Everything Allocated
Go

91. Memory Locality
Java
No value types
Everything Allocated
Go
Structs
True Value types

92. Memory Locality
Java
No value types
Everything Allocated
Can’t return multiple values
Go
Structs
True Value types

93. 20172000 2005 201090s80s
Software|
Languages
Hardware&
Compute
Context
Go
Memory
Locality
C++ at
Google
C10K
Epoll
(linux)
Kqueue
(BSD)
SEDA
Java NIO
P4
Core 2
Ruby
Mongrel
Gunicorn
(Python)
NodeJS
The rise of
containers +
cloud-native+
ecosystem
Microservices
serverless
Hotspot
JVM
Go GC
1.8

94. When the three of us [Ken Thompson, Rob Pike, and Robert
Griesemer] got started, it was pure research. The three of us got
together and decided that we hated C++. [laughter] ... [Returning
to Go,] we started off with the idea that all three of us had to be
talked into every feature in the language, so there was no
extraneous garbage put into the language for any reason.

95. 20172000 2005 201090s80s
Software|
Languages
Hardware&
Compute
Context
Go
Memory
Locality
C++ at
Google
C10K
Epoll
(linux)
Kqueue
(BSD)
SEDA
Java NIO
P4
Core 2
Ruby
Mongrel
Gunicorn
(Python)
NodeJS
The rise of
containers +
cloud-native+
ecosystem
Microservices
serverless
Hotspot
JVM
Go GC
1.8

96. Memory Locality
Java
No value types
Everything Allocated
Can’t return multiple values
Go
Structs
True Value types

97. Memory Locality
Java
No value types
Everything Allocated
Can’t return multiple values
Go
Structs
True Value types
compact object layout
No object headers

98. Memory Locality
Java
No value types
Everything Allocated
Can’t return multiple values
Go
Structs
True Value types
compact object layout
No object headers
UTF-8

99. Memory Locality
Java
No value types
Everything Allocated
Can’t return multiple values
Go
Structs
True Value types
compact object layout
No object headers
UTF-8UTF-16

100. Memory Locality
Java
No value types
Everything Allocated
Can’t return multiple values
Go
Structs
True Value types
Compact object layout
No object headers
Lazy initialization of
collections
UTF-8UTF-16

101. Memory Locality (conclusion)

102. Memory Locality (conclusion)
● Go gives programmers the tools to talk about memory
efficiently if they need it.

103. Memory Locality (conclusion)
● Go gives programmers the tools to talk about memory
efficiently if they need it.
● Flexible

104. Memory Locality (conclusion)
● Go gives programmers the tools to talk about memory
efficiently if they need it.
● Flexible
● Memory management (not an all-or-nothing like in C++ or
Rust)

105. Readability

106. Readability
“ Debugging is twice as hard as writing the code in the first
place. Therefore, if you write the code as cleverly as possible,
you are, by definition, not smart enough to debug it. ” — Brian
Kernighan

107. 20172000 2005 201090s80s
Software|
Languages
Hardware&
Compute
Context
Go
Memory
Locality
C++ at
Google
C10K
Epoll
(linux)
Kqueue
(BSD)
SEDA
Java NIO
P4
Core 2
Ruby
Mongrel
Gunicorn
(Python)
NodeJS
The rise of
containers +
cloud-native+
ecosystem
Microservices
serverless
Hotspot
JVM
Go GC
1.8

108. 20172000 2005 201090s80s
Software|
Languages
Hardware&
Compute
Context
Go
Memory
Locality
C++ at
Google
C10K
Epoll
(linux)
Kqueue
(BSD)
SEDA
Java NIO
P4
Core 2
Ruby
Mongrel
Gunicorn
(Python)
NodeJS
The rise of
containers +
cloud-native+
ecosystem
Microservices
serverless
Hotspot
JVM
Go GC
1.8
# of google
employees in
2007: 16,805

109. 20172000 2005 201090s80s
Software|
Languages
Hardware&
Compute
Context
Go
Memory
Locality
C++ at
Google
C10K
Epoll
(linux)
Kqueue
(BSD)
SEDA
Java NIO
P4
Core 2
Ruby
Mongrel
NodeJS
The rise of
containers +
cloud-native+
ecosystem
Microservices
serverless
Hotspot
JVM
Go GC
1.8
2010:
24,440

110. 20172000 2005 201090s80s
Software|
Languages
Hardware&
Compute
Context
Go
Memory
Locality
C++ at
Google
C10K
Epoll
(linux)
Kqueue
(BSD)
SEDA
Java NIO
P4
Core 2
Ruby
Mongrel
Gunicorn
(Python)
NodeJS
Hotspot
JVM
Go GC
1.8
2017:
75,606

111. Readability
simplicity

112. Readability
simplicity
“simple is better”

113. Readability
simplicity
“simple is better”
“this is an insult to
intelligent programmers”

114. Readability
simplicity
“simple is better”
“you’re trying to
commodify programming
and create a situation
where our bosses can
replace us at will”

115. “You’re not paid to program, you’re not even paid to maintain
someone else’s program, you’re paid to deliver solutions to the
business.”
- Dave Cheney

116. Readability
Programs which cannot be maintained will be rewritten

117. Readability
Programs which cannot be maintained will be rewritten
“If you can’t be replaced, you cannot be promoted”

118. 20172000 2005 201090s80s
Software|
Languages
Hardware&
Compute
Context
Go
Memory
Locality
C++ at
Google
C10K
Epoll
(linux)
Kqueue
(BSD)
SEDA
Java NIO
P4
Core 2
Ruby
Mongrel
Gunicorn
(Python)
NodeJS
Hotspot
JVM
Go GC
1.8
2017:
75,606
Average length of tenure at company:

119. 20172000 2005 201090s80s
Software|
Languages
Hardware&
Compute
Context
Go
Memory
Locality
C++ at
Google
C10K
Epoll
(linux)
Kqueue
(BSD)
SEDA
Java NIO
P4
Core 2
Ruby
Mongrel
Gunicorn
(Python)
NodeJS
Hotspot
JVM
Go GC
1.8
2017:
75,606
Average length of tenure at company:

120. 20172000 2005 201090s80s
Software|
Languages
Hardware&
Compute
Context
Go
Memory
Locality
C++ at
Google
C10K
Epoll
(linux)
Kqueue
(BSD)
SEDA
Java NIO
P4
Core 2
Ruby
Mongrel
Gunicorn
(Python)
NodeJS
Hotspot
JVM
Go GC
1.8
2017:
75,606
Average length of tenure at company:

121. 20172000 2005 201090s80s
Software|
Languages
Hardware&
Compute
Context
Go
Memory
Locality
C++ at
Google
C10K
Epoll
(linux)
Kqueue
(BSD)
SEDA
Java NIO
P4
Core 2
Ruby
Mongrel
Gunicorn
(Python)
NodeJS
Hotspot
JVM
Go GC
1.8
2017:
75,606
Average length of tenure at company:

122. Software Engineering

123. Software Engineering
Software Engineering vs Programming

124. Software Engineering
Software Engineering vs Programming
Software Engineering = Programming integrated over time.

125. Software Engineering
Software Engineering vs Programming
Software Engineering = Programming integrated over time.
Engineering is what happens when things need to live longer and influence of
time starts creeping in. -Titus Winters

126. Software Engineering
Software Engineering vs Programming
Software Engineering = Programming integrated over time.
Engineering is what happens when things need to live longer and influence of
time starts creeping in. -Titus Winters
All this complexity is fundamentally a different flavor than programming.

127. Software Engineering
focus on sustaining engineering (readability)

128. Software Engineering
focus on sustaining engineering (readability)
continuance of many different engineers over a long period of time

129. Software Engineering
focus on sustaining engineering (readability)
continuance of many different engineers over a long period of time
clear module boundaries

130. Software Engineering
focus on sustaining engineering (readability)
continuance of many different engineers over a long period of time
clear module boundaries
keeping import dependencies between packages linear, thus keeping compile
times down.

131. Simplicity and the Greater Good

133. “Simplicity is a great virtue but it requires hard work to achieve it
and education to appreciate it. And to make matters worse:
complexity sells better.”
― Edsger W. Dijkstra

134. The Future

135. 20172000 2005 201090s80s
Software|
Languages
Hardware&
Compute
Context
Go
C++ at
Google
C10K
Epoll
(linux)
Kqueue
(BSD)
SEDA
Java NIO
P4
Core 2
Ruby
Mongrel
Gunicorn
(Python)
NodeJS
Hotspot
JVM
Go GC
1.8
2010:
75,606
Average length of tenure at company:

136. Software|
Languages
Hardware&
Compute
Context
20402025 2030 20352017 2020
The Future?

137. Software|
Languages
Hardware&
Compute
Context
20402025 2030 20352017 2020
The Future?
The problems we have today were not there 20
years ago, nor will be problems we face 20 years
from now.

138. Software|
Languages
Hardware&
Compute
Context
20402025 2030 20352017 2020
The Future?
...it may surprise you

139. Thank you!
Carmen Andoh @carmatrocity
QCon San Francisco
21st Century Languages Track
November 2017

140. Watch the video with slide
synchronization on InfoQ.com!
https://www.infoq.com/presentations/
go-concurrency-gc